OpenAI has updated its usage policies to require mandatory review and approval for model routers and AI gateways. These platforms, which act as intermediaries between developers and AI models, must obtain explicit authorization from OpenAI before offering third-party access to its technology.
The new requirements take effect on November 20, 2026. The policy revision was officially dated October 21, 2026, providing operators with approximately one month to comply with the updated standards.
The policy specifically targets platforms that resell direct access to OpenAI models. It does not apply to applications that utilize OpenAI models to power their own internal features and products.
The update is part of a broader overhaul of OpenAI’s usage policies, which also includes clarified rules regarding the use of model outputs for training other AI systems, protections for minors, restrictions on high-stakes use cases, and new national-security requirements.
