‘We may release models we can’t control’: Amodei, Altman … and Gemini Carbon
The discussion covers the shift in AI safety policies, the challenges of controlling frontier models, and the technical limits of human understanding when dealing with complex LLM architectures.
Read article