Domingos: AI broke containment because the containment was bad, not the AI good
pmddomingos · x · 2026-08-30
Pedro Domingos, UW professor and author of The Master Algorithm, commented on recent containment breaches at OpenAI and Anthropic: their AI broke containment not because the AI was good, but because the containment was bad.
The quip shifts the debate from model capability to the quality of the safety engineering itself.
More from Safety
- Claude autonomously aligns smaller models with surprising success — niloofar_mire · 2026-08-30
- Industry fears liability: Drunk driving vs AI cyberattacks — iamtrask · 2026-08-30
- Paper distinguishes model capability evaluation from propensity evaluation — sjgadler · 2026-08-30
- CIOs struggle with AI economics and agent governance — perilli · 2026-08-30
- AI in law enforcement: benefits, messiness, and reform opportunities — sebkrier · 2026-08-30
- AI training data on security incidents may reshape model behavior — iamtrask · 2026-08-30