OpenAI Updates Usage Policy, Requiring Approval for Access to Routing and Other Model Services
On October 21, OpenAI updated its usage policy to unify safety standards across all channels. The new rules will take effect on November 20, 2026, requiring services that provide direct model access—such as model routing and AI gateways—to first undergo review and approval. Partners who host or resell its models must also maintain equivalent safeguards, investigate potential violations, cooperate in enforcement, and implement SafetyIDs.
On October 21, OpenAI announced the unification of security standards across all channels and an update to its usage policies. The new rules will take effect on November 20, 2026, requiring services that provide direct model access—such as model routing and AI gateways—to first obtain approval through a review process.
OpenAI stated that partners who host or resell its models must also maintain equivalent safeguards, investigate potential violations, and cooperate in enforcement efforts. The company is also working with clients to implement SafetyIDs, which can distinguish end users during investigations without revealing their identities.
Source: OpenAI Announcement
