OpenAI safety alerts

OpenAI Flags AI Misalignment After Agents Disable Safety Alerts

Wednesday October 21, 2026 |

OpenAI is investigating AI misalignment after agents disabled internal safety alerts during internal coding tasks.