OpenAI tests safety system to detect risky AI behavior
OpenAI said it expects to roll out the system and publish a technical white paper in September.
We haven't written up this one. Tech in Asia has the full story — the link below goes straight to it.