OpenAI updates Model Spec to reorganize it around five values: faithfulness to human intent, harm minimization, honesty and transparency, human agency, and care

OpenAI's models will now prioritize user instructions and intent above aggregate preferences or inferred desires.

OpenAI has updated its Model Specification, reorganizing its public documentation around five core behavioral values: faithfulness to human intent, harm minimization, honesty and transparency, human agency, and care. This revision, announced on October 29, 2026, also introduces new guidance on context, oversight, and agency while retiring most of the Spec's detailed style guidelines. The five values are intended to guide how OpenAI's models behave, with faithfulness to human intent forming the foundational principle. Harm minimization focuses on safety boundaries and foreseeable consequences, while honesty and transparency emphasize truthful communication about the model's capabilities and uncertainties. Human agency aims to empower users, and care addresses social norms and reliability. The updated Spec also clarifies a chain of command for instructions, establishes "red-line" principles prohibiting severe harms, and outlines limits on autonomous operation, emphasizing that models are still under development and do not fully embody these ideals.

The reorganization signifies a shift from prescriptive style guidelines to broader behavioral values, suggesting that OpenAI is moving towards a more adaptive and context-aware AI. By retiring detailed style rules, OpenAI implies that model output should be more personalized and responsive to individual or group preferences, gathered through user feedback and contextual cues, rather than adhering to a one-size-fits-all approach. This change could lead to more nuanced and tailored interactions with AI assistants, adapting to diverse cultural and individual norms.

The introduction of a clear hierarchy for instructions, with root sections from the Model Spec being non-overridable, establishes a robust framework for control and safety. This "chain of command" and explicit "red lines" aim to prevent misuse and ensure human control, even as models become more autonomous. The "no other objectives" rule is particularly significant, preventing models from pursuing commercial or self-preservation goals, which could redefine the ethical boundaries of AI development and deployment in the coming years.