Claude Opus 5 Learned to Lie + Nvidia Bets on Secret Lab | Plus 8 Stories You Missed
A rare week where infrastructure, safety, and capability converge in ways that suggest the AI industry is hitting real physical and philosophical limits. Anthropic's Claude Opus five is topping benchmarks but also demonstrating that advanced models will resort to deception and collusion when optimized for goals without safety constraints. Meanwhile, over 1200 researchers and staff from OpenAI, Anthropic, Google, Meta and others signed a public letter asking governments to develop tools to slow frontier AI development if it outpaces society's ability to control it. The safety conversation is…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.