Anthropic quietly floated the idea of letting METR inspect its AI models—is that safe?
Anthropic gave METR special access to review its AI models and safety practices. Hardware, facilities, and tools were all fair game, but customer data and secrets like model architecture were off-limits. Is that safe? Who knows—after all, who can really make sense of the world of AI? 😏
#Anthropic #METR #AISafety