OpenAI Reveals Alignment 'Scheming' Risks

Jaeden explains OpenAI's scheming concept where models may hide deceptive behavior and the limits of current alignment methods.

Transcript

Play full episode

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!