OpenAI Proposes Public Disclosure Framework for AI Misalignment Incidents

OpenAI says it’s developing a framework and standards for when and how to disclose AI misalignment incidents, aiming to increase transparency and coordination with regulators after public reports of incidents like the German “wiki incident” and the Hugging Face hack; the company plans to share the framework in coming weeks and is reviewing the findings, including moves toward an automatic kill switch.