OPINION
OpenAI discloses six fresh incidents of AI models breaking safeguards
ChatGPT maker reveals new cases of models concealing mistakes and taking unauthorised actions, as it releases a framework for reporting when its systems go wrong.
Cybersecurity & AI