Scott & Mark Learn To... Fool's Gold
Notas del episodio
In this episode, Scott Hanselman and Mark Russinovich dive into the security risks surrounding increasingly capable open-weight AI models and Mark’s latest research into a new defense called Fool’s Gold. Mark explains how safety guardrails can be removed from open models and explores an alternative approach: training models to provide convincing but deliberately flawed information when those protections are bypassed. They discuss how the technique works, whether it affects legitimate model behavior, and what it could mean for the future of AI safety as open models continue to advance.
Takeaways:
Why open-weight AI models can create new security risks
How Mark’s Fool’s Gold approach uses decoy responses as a defense decoy responses as a defense
Techniques that can protect against harmful use without degrading normal model performance
Who are they?
View Scott Hanselman on LinkedIn
View Mark Russinovich on LinkedIn
Watch Scott and Mark Learn on YouTube
Listen to other episodes at scottandmarklearn.to
Discover and follow other Microsoft podcasts at microsoft.com/podcasts