Think Different. Think AI. Transcript archive

AI Security

Prompt Injection, voice cloning, exploits built from security patches: what happens when you give a model permissions, and what attacks on it look like in practice.