# An updated Model Spec for more capable and beneficial AI

The revised behavioral guidelines center on faithfulness to human intent, harm minimization, honesty, human agency, and care while retiring detailed style prescriptions.

- The updated chain of command details how instructions apply and resolve conflicts between platform rules, developers, and users
- A new section guides models to draw on conversation history, memories, and context to infer user intent
- Detailed style guidance has been largely retired, letting style adapt to user feedback and context while retaining an objective default perspective
- The specification document now refers to the system as GPT rather than the assistant across deployment settings

## Why it matters

Developers and users building on OpenAI's models can see the explicit rules and defaults that govern how future model training interprets instructions and handles behavioral trade-offs.

## Sources

- [OpenAI: An updated Model Spec for more capable and beneficial AI](https://openai.com/index/updated-model-spec)

---

Summarized by dstilled on 2026-10-29. https://dstilled.ai/story/64ca2608-f08d-4b14-804c-2dd31e008cf2
