OpenAI updates its Model Spec with five core values and a new name for its AI

The revised rulebook swaps "the assistant" for "GPT," drops most style rules, and leans harder on human oversight OpenAI has rewritten the rulebook for how its AI models are supposed to behave. On October 29, 2026, the company released an updated version of its Model Spec built around five core values. The update also renames the subject of the document. Throughout the text, “the assistant” is now simply “GPT.” What changed in the Model Spec The Model Spec is a public document that lays out how OpenAI expects its models to act. The new version organizes everything around five values. These are faithfulness to human intent, harm minimization, honesty and transparency, human agency, and care. The revision adds explicit principles covering user agency, social norms, and support for oversight. One new piece of guidance is labeled “Consider context carefully.” According to the research summary, those principles are meant to allow more context-aware personalization while users keep their privacy and control. OpenAI removed most of the detailed style guidance that earlier versions included. Instead of prescribing exactly how the model should phrase things, the new approach lets communicat...