An updated Model Spec for more capable and beneficial AI An updated Model Spec for more capable and beneficial AI openai.com/ai 2026-10-29 1 min read original #safety OpenAI’s updated Model Spec outlines five core values to guide more capable AI toward sound judgment, human agency, and faithfulness to human intent.