OpenAI to require approval for model routers
OpenAI will require model routers and AI gateways to get its review and approval before offering its models from November 20, a change that tightens control over third-party distribution channels.
OpenAI announced updated Usage Policies, effective November 20, 2026. Under the new rules, services that provide direct access to its models, including model routers and AI gateways, must get OpenAI's review and approval before offering that access.
The policies cover every route to OpenAI's models: OpenAI's own services, partners that host the models, and resellers. Partners must keep safeguards consistent with OpenAI's own, investigate potential violations, and take action when they find problems. They must also coordinate with OpenAI on enforcement so that a restriction imposed through one service still holds on the others. OpenAI said it has spent several months working with customers and partners to align safety policies, and it cited models such as GPT-6 Astra as the reason consistent safeguards matter more as capabilities grow.
OpenAI is also introducing SafetyIDs with customers. Customers can supply these identifiers so that OpenAI can tell individual end users apart during a misuse investigation without learning who they are. The company said this lets it target its response more precisely and cause less disruption to other users. It also mentioned Private Safety Processing as an example of safeguards that give customers more control over sensitive data, but it did not explain how that system works.
OpenAI said its enforcement decisions are based on documented conduct and risk and are proportionate to the circumstances. It said it will revisit restrictions if conduct, commitments, or safeguards change meaningfully.
The post does not say what the approval process involves, how long it takes, or what happens to existing routers that are not approved by the deadline.