{"id":131272,"date":"2023-08-07T08:00:00","date_gmt":"2023-08-07T15:00:00","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/security\/blog\/?p=131272"},"modified":"2024-07-03T07:56:35","modified_gmt":"2024-07-03T14:56:35","slug":"microsoft-ai-red-team-building-future-of-safer-ai","status":"publish","type":"post","link":"https:\/\/www.microsoft.com\/en-us\/security\/blog\/2023\/08\/07\/microsoft-ai-red-team-building-future-of-safer-ai\/","title":{"rendered":"Microsoft AI Red Team building future of safer AI"},"content":{"rendered":"\n
An essential part of shipping software securely is red teaming. It broadly refers to the practice of emulating real-world adversaries and their tools, tactics, and procedures to identify risks, uncover blind spots, validate assumptions, and improve the overall security posture of systems. Microsoft has a rich history<\/a> of red teaming emerging technology with a goal of proactively identifying failures in the technology. As AI systems became more prevalent, in 2018, Microsoft established the AI Red Team: a group of interdisciplinary experts dedicated to thinking like attackers and probing AI systems for failures.<\/p>\n\n\n\n We\u2019re sharing best practices from our team so others can benefit from Microsoft\u2019s learnings. These best practices can help security teams proactively hunt for failures in AI systems, define a defense-in-depth approach, and create a plan to evolve and grow your security posture as generative AI systems evolve.<\/p>\n\n\n\n The practice of AI red teaming has evolved to take on a more expanded meaning: it not only covers probing for security vulnerabilities, but also includes probing for other system failures, such as the generation of potentially harmful content. AI systems come with new risks, and red teaming is core to understanding those novel risks, such as prompt injection and producing ungrounded content. AI red teaming is not just a nice to have at Microsoft; it is a cornerstone to responsible AI by design: as Microsoft President and Vice Chair, Brad Smith, announced, Microsoft recently<\/a> committed that all high-risk AI systems will go through independent red teaming before deployment. <\/p>\n\n\n\n The goal of this blog is to contextualize for security professionals how AI red teaming intersects with traditional red teaming, and where it differs. This, we hope, will empower more organizations to red team their own AI systems as well as provide insights into leveraging their existing traditional red teams and AI teams better.<\/p>\n\n\n\n Over the last several years, Microsoft\u2019s AI Red Team has continuously created and shared content to empower security professionals to think comprehensively and proactively about how to implement AI securely. In October 2020, Microsoft collaborated with MITRE as well as industry and academic partners to develop and release the Adversarial Machine Learning Threat Matrix,<\/a> a framework for empowering security analysts to detect, respond, and remediate threats. Also in 2020, we created and open sourced Microsoft Counterfit<\/a>, an automation tool for security testing AI systems to help the whole industry improve the security of AI solutions. Following that, we released the AI security risk assessment framework<\/a> in 2021 to help organizations mature their security practices around the security of AI systems, in addition to updating Counterfit. Earlier this year, we announced<\/a> additional collaborations with key partners to help organizations understand the risks associated with AI systems so that organizations can use them safely, including the integration of Counterfit into MITRE tooling, and collaborations with Hugging Face on an AI-specific security scanner that is available on GitHub.<\/p>\n\n\nRed teaming helps make AI implementation safer<\/h2>\n\n\n\n