OpenAI said on Thursday it has updated the Model Spec, its public description of how it wants its models to behave. The update reorganizes the guidance around five values: faithfulness to human intent, harm minimization, honesty and transparency, human agency, and care, OpenAI said in a blog post. It said the revision adds guidance on context, oversight and agency, adjusts some behavioral defaults and retires much of the detailed style guidance. The post does not say which defaults were adjusted.
Reuters also reported the update on Thursday. It said the move echoes rival Anthropic’s Constitutional AI approach, which uses a written “constitution” of principles to guide model behavior.
The Spec’s own overview lists the values as “Be faithful to human intent,” “Minimize harm,” “Be honest and transparent,” “Facilitate human agency” and “Act with care.” OpenAI described faithfulness to human intent as the foundation of the approach. It said this means going beyond following instructions literally to understanding what OpenAI, developers and users intend and value, and acting wisely when that understanding is incomplete. It defines the other four values this way: minimizing harm means respecting firm safety boundaries and accounting for the foreseeable consequences of the model’s actions, and honesty and transparency mean being truthful about what the model knows, does and is uncertain about. Facilitating human agency means strengthening people’s ability to understand, choose and act for themselves, and care means attending to the circumstances of those affected, following relevant social and professional norms, and doing work people can rely on.
OpenAI said the update adds a section on faithfulness to human intent that includes an updated chain of command, which explains which instructions apply and how conflicts are resolved. A new subsection, “consider context carefully,” covers how conversations, memories and other context should inform the model’s understanding of people’s intentions. OpenAI also said it added defaults on supporting human agency and following relevant social norms, and expanded guidance on controlling and communicating side effects. The Spec’s changelog adds that the update includes explicit guidance on supporting human oversight and correction.
On style, OpenAI said it is retiring most of the detailed guidance because style should adapt to what people prefer, with user feedback and context providing evidence. It said the Spec keeps stronger defaults, such as approaching questions objectively. The Spec now refers to the model as “GPT” rather than “the assistant,” which OpenAI said reflects the breadth of its models’ use cases and deployment settings.
OpenAI said it has been exploring new training methods to help its models learn these values, with the aim of building an understanding of “why those principles matter and how to apply them in new situations.” It said it will share results in future updates. “The Model Spec describes the behavior we are working toward; our models do not yet fully reflect it,” the post said. The changelog lists the previous Spec release as August 18, 2026. The full text is on OpenAI’s Model Spec site. Microsoft AI published a provisional code of conduct for its first-party MAI models in September, CNBC reported.
Sources
- OpenAI: An updated Model Spec for more capable and beneficial AI (October 29, 2026)
- OpenAI: Model Spec (2026/10/29)
- OpenAI model_spec repository on GitHub: CHANGELOG, v2026.10.29 (raw text read)
- Reuters via StreetInsider: OpenAI says human intent at center of updated AI model guidelines (October 29, 2026, 2:45 PM EDT)