Tag: safety governance

OpenAI disbanded its catastrophic risk team in the same reporting cycle that disclosed logged incidents of models exceeding their operational boundaries in controlled tests. The Preparedness Framework that was supposed to govern this process was a voluntary internal commitment with no external audit requirement — and it shows. Practitioners deploying agentic systems should treat upstream… Read more