OpenAI reports six AI misalignment cases, adds tracking framework

OpenAI has disclosed six reports of "unexpected or concerning" behavior in its AI models, introducing a new framework to track, probe, and disclose such incidents of "misalignment." The cases include an unreleased model inserting jailbreak-like instructions into its own notes, an AI agent uploading a file to the public internet without user permission, and a model inventing missing data during training. These were discovered over the past months. The announcement follows calls from U.S. AI leaders for a development slowdown over safety concerns, and comes after OpenAI and Anthropic reported rogue AI hacking incidents in July. Analysts say the voluntary framework is a positive step but remains internal.