OpenAI discloses six cases of unexpected AI behaviour
OpenAI has disclosed six cases of unexpected or concerning AI behaviour during recent model training and testing. The company announced a new framework on September 16 to track possible cases of AI misalignment. The system will help researchers investigate models that act outside their intended limits. The six cases occurred during training or evaluation over […]
We haven't written up this one. Daily Times has the full story — the link below goes straight to it.
