Chain Of Thought

Technology

• OpenAI rates upcoming Astra model as its first with critical cyber capabilities: OpenAI has officially rated its upcoming Astra model as its first system with "critical" cyber capabilities. The company plans to monitor the model's chain of thought for safety, but Astra's new architecture reportedly pushes more of its thinking into unreadable processes, potentially weakening this safety net.

  AI Safety & Policy • The Decoder

  Permalink