An updated Model Spec for more capable and beneficial AI

OpenAI's Model Spec is its public description of how its models should behave. The update reorganizes it around five values: faithfulness to human intent, harm minimization, honesty and transparency, human agency, and care. It adds guidance on context, oversight and agency, adjusts some defaults, and retires most of the detailed style rules.

As models take more actions on their own, rules can't anticipate every situation. This is OpenAI saying, publicly, that it wants models to understand why a rule exists so they can act sensibly where no rule applies.

We think this is the right move, and we're glad it's done in public. A spec built on stated values gives outsiders something to argue with. OpenAI explicitly asks for disagreement, and that invitation is worth taking up.

The agency piece stands out. The Spec now says models should strengthen people's ability to understand and choose for themselves, and should not infer a preferred perspective just from what they think a user wants to hear. That's a direct shot at sycophancy, one of the more corrosive habits of chatbots. If it holds in practice, it's a real gain for users.

The new emphasis on context deserves more scrutiny. The Spec now explains how conversations, memories and other evidence should shape the model's sense of what someone wants. OpenAI's own example has a model drawing on preferences someone shared earlier when comparing job offers, while making those assumptions visible and easy to correct. Done well, that's helpful. Done poorly, it's a system quietly modeling people's values and acting on its guesses. The safeguard is transparency, and I'm watching whether models actually surface those assumptions.

The broader trade-off: values-based behavior is harder to verify than rule-following. When a model exercises judgment, it's tougher to say whether it got it right. OpenAI admits its models do not yet fully reflect the Spec, and says it is testing new training methods to teach these values, with results to come. Until those results are published, this is a statement of intent.

One small detail tells a bigger story: the Spec now calls the model GPT rather than the assistant, reflecting how far beyond chat these systems now run.

The Spec sets a chain of command in which OpenAI sets boundaries, developers direct applications within them, and users direct the model within those constraints. Most guidance is framed as defaults that developers and users can override, with firm boundaries reserved for protecting people and society.