The Generative AI Security Podcast cover image

Are Your Red Teaming Efforts Giving Bad Actors An Advantage? GenAI Security

The Generative AI Security Podcast

00:00

Limits and Risks of Red Teaming for Evasion

Susanna presents mathematical limits on achieving full evasion-test coverage, explains how published red-team tests can advantage attackers, discusses QA vs production testing risks, and previews a new evasion-testing methodology and forthcoming paper.

Play episode from 08:44
Transcript

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!
App store bannerPlay store banner
Get the app