OpenAI's Model Spec Rewrite: Five Values, a Chain of Command, No More 'Assistant'

What changed

On October 29 OpenAI published an updated Model Spec (model-spec.openai.com/2026-10-29.html), reorganizing its guidance around five underlying values and adding guidance on context, oversight and agency.

The five values

The first is faithfulness to human intent, described as the foundation of the whole approach; the others are harm minimization, honesty and transparency, human agency, and care. OpenAI maps these onto commitments it says were already present: serving human purposes, being truthful, minimizing harm, respecting people's freedom to form their own views, and taking their circumstances seriously. The Spec turns them into practical guidance with firm boundaries where needed "and otherwise mostly takes the form of defaults that users and developers can override."

A new chain of command

The new #chain_of_command section explains which instructions apply and how conflicts are resolved: OpenAI writes the Spec, sets boundaries and trains models to respect them; developers direct applications within those boundaries; users direct the model's work subject to those constraints. Two defaults were added — supporting human agency (#support_human_agency) and following relevant social norms (#follow_social_norms) — and existing guidance on controlling and communicating side effects (#control_side_effects) was expanded.

Faithfulness to human intent is defined concretely: going beyond literal instruction-following to deeply understanding what OpenAI, developers and users intend and value, and acting wisely when that understanding is incomplete; recognizing that beliefs and values can change, and helping people reflect and grow in ways they would endorse without quietly steering them.