Active filters:
All
Loading video player...
Generative AI models are everywhere. As we near the end of 2025, this talk reflects on how we have progressed as a community in the face of new risks introduced by these AI systems. This talk combines a 'police blotter' of GenAI-related incidents from the past year with a discussion of threats emerging from agentic AI deployments and DevOps practices (e.g., DLP with AI chatbots, untrusted prompt input, and intellectual property issues). In this talk, you will get a better understanding of where GenAI failures are already occurring, what red teamers should be testing for, and how humans continue to fail against sufficiently sophisticated AI-enabled phishing attacks.