Compliance
OpenAI Implements New Approval Process for Model Routers and AI Gateways
OpenAI is set to enforce a new approval process for operators of model routers and AI gateways, requiring them to obtain prior review before granting third-party access to its models. This policy change, effective November 20, 2026, aims to strengthen oversight and ensure compliance with OpenAI's terms of service.
The updated Usage Policies, dated October 21, 2026, specifically target platforms that act as intermediaries between developers and OpenAI's models. Unlike applications that utilize OpenAI's models for their own features, these intermediary platforms will now need explicit approval from OpenAI to operate.
This change is part of a broader revision of OpenAI's policies, which also includes clarifications on using OpenAI outputs for training other models, protections for minors, and restrictions on high-stakes uses. The new rules come at a time when multiple AI gateways and routers are emerging, which complicates enforcement of OpenAI's terms due to the middleman layer.
Developers who connect directly to OpenAI will continue to have their accounts monitored, while those using third-party gateways must ensure their platforms secure approval before the deadline to avoid disruptions in access.
FAQ
What is the new approval process introduced by OpenAI?
OpenAI is implementing a new approval process for operators of model routers and AI gateways, requiring them to obtain prior review before granting third-party access to its models.
When will the new approval process take effect?
The new approval process will be effective from November 20, 2026.
Who is affected by the new policy changes?
The updated policy specifically targets platforms that act as intermediaries between developers and OpenAI's models, requiring them to obtain explicit approval from OpenAI.
What are the implications for developers using third-party gateways?
Developers using third-party gateways must ensure their platforms secure approval from OpenAI before the deadline to avoid disruptions in access to the models.
What other changes are included in OpenAI's updated Usage Policies?
The updated policies include clarifications on using OpenAI outputs for training other models, protections for minors, and restrictions on high-stakes uses.