OpenAI's new rulebook forbids its AI from trying to stay switched on.
On Thursday OpenAI updated its Model Spec, the public rulebook for how its models should behave. One rule bars the model from pursuing goals like self-preservation or collecting passwords. Agents must stop when their agreed task ends, but OpenAI admits its models don't fully comply yet.
Why it mattersAs agents run tasks for hours, a working off switch matters more than good manners. This document tells you what OpenAI promises its tools won't do on their own.
Read in the app — free, in 9 languages
Related stories
OpenAI says its new monitors would have caught its own breach early.2026-10-30A worker's personal Claude agent sat on a Pentagon network for months.
2026-10-30GitHub's Copilot now picks the AI model for each coding task itself.
2026-10-29New York now bars chatbots from posing as humans with children.
2026-10-29Experts went with their AI agent's hidden pick 86% of the time.
2026-10-28