Probing the Limits: A Workshop on Red-Teaming AI Systems

This workshop will be held at the Software Engineering Institute on January 28, 2025. Sessions will be recorded.

Program chairs: Anusha Sinha, Lauren McIlvenny, and Nathan VanHoudnos

Cyber red teaming is regularly employed to test systems for weak points before an adversary discovers and exploits them. Red-teaming is also increasingly employed to evaluate the safety risks of Artificial Intelligence (AI) systems given the broad risk surface of modern models, but there is a lack of clear guidance on how to effectively conduct red-teaming for AI. Given the widespread adoption of Generative AI, the SEI is working to identify best practices for red teaming Generative AI Systems relative to existing guidance on cyber red teaming. We hope that you will join us as we discuss our findings and collaborate to codify best practices for the future.

Host or Publisher Software Engineering Institute at Carnegie Mellon University