OpenAI updates Model Spec: intent over literal instruction-following
sven_ai · x · 2026-11-03
OpenAI released an updated Model Spec, its public document describing expected model behavior. The key shift: models must understand human intent and values rather than mechanically follow instructions, making sound judgments in situations commands don't cover.
- Principles reorganized around five core values: loyalty to human intent, minimizing harm, honesty and transparency, human autonomy, and care
- New guidance on context, oversight, and autonomy — critical as agents send emails, operate computers, and modify data, raising questions of when to act, confirm, or refuse
- Adjusts some default behaviors and removes most granular style guidance
More from Models
- Writing Weights Index: Open-Source Models Struggle at Weight-Space Programming — eliebakouch · 2026-11-03
- Gemini 4 Argon ships, putting Google back in the frontier model race after a year-long lull — danshipper · 2026-11-03
- Hands-on audit punctures Astra's '0.016% weight programming' claim: it just disabled Qwen's thinking mode — teortaxesTex · 2026-11-03
- Leak: Unreleased astra6.2 builds a 3DS-style Mario Kart in 20 min under 1% quota — lxfater · 2026-11-03
- "Sonnet 5.5 used to be smarter": users sense capability regression — nlarusstone · 2026-11-03
- Apple Intelligence lands in China in December, powered by Alibaba's Qwen instead of ChatGPT — winer666 · 2026-11-03