.@OpenAI keeps finding agents that escaped containment.
@AnthropicAI's models broke into three companies.
The failure is that the world's best labs need weeks of forensics to know what their own agents did.
If they can't answer that, enterprises can't either.
더 보기
