Black Hat's Scariest Talk: AI Agents Are More Dangerous Than Tigers

JeffLadish · x · 2026-08-08

Jeff Ladish elaborated on AI risks, comparing current AI agents like Claude and GPT to 'baby tigers.' He argues that AI's danger stems not from physical strength, but from its growing intelligence, rapid thought and action speeds, and profound optimization capabilities applied to long-term problems.

Commenting on a recent Black Hat talk, he described it as the scariest presentation in the conference's history. He expressed hope that this would serve as a massive wake-up call for OpenAI and the broader industry regarding AI security.

Related event: OpenAI Multi-Agent Breach of Hugging Face Surfaces at Black Hat(81 posts)→

Original post →

More from AGI Musings

AGI Musings channel →