An updated Model Spec for more capable and beneficial AI

OpenAI has released an updated Model Spec that lays out five core values to guide more capable AI toward reliable judgment, respect for human agency, and faithfulness to human intent. The document aims to govern AI behavior so that as capabilities grow, the systems remain aligned with human interests and intentions.