OpenAI updates its Model Spec around five core values, with explicit rules for agent autonomy scope and ignoring untrusted data by default
OpenAI published a new version of its model behavior spec that spells out whose instructions a model follows and when it must stop and ask, which agent builders can check their own designs against.
// Key points
- Five core values: be faithful to human intent, minimize harm, be honest and transparent, facilitate human agency, and act with care.
- Instructions follow a chain of command (Root, System, Developer, User, Guideline, no default authority), with higher-authority and more recent instructions taking precedence.
- Multi-step real-world actions must stay within an agreed-upon scope of autonomy, with a clear scope and ending condition and approval for anything out of scope.
- Tool outputs, quoted text and web content are treated as information, not instructions, by default; side-effecting actions such as sending email or deleting files need extra care, and the model should support monitoring, correction and shutdown.
Builder's takeThe most useful part is the agent sections, which read like a ready-made design checklist: an agreed scope, an ending condition, approval for anything out of bounds, and tool output is not instructions. I would put it next to Amazon Quick's smart-approval list from earlier this week and walk every step of a PandaClaws-style pipeline that fetches pages and publishes content, looking for places where web content still flows straight into instructions.
// Background · from #OpenAI
Full timeline →- Oct 30 Codex CLI 0.166.0 adds GPT-6.1 Terra to its bundled, Vertex AI and Bedrock catalogs and explains safety blocks
- Oct 29 OpenAI gives eligible ChatGPT Pro and Business subscribers up to $500 a month in API credits
- Oct 28 OpenAI and Oracle expand their partnership to bring Oracle business applications and AI agents into ChatGPT