From the source
OpenAI is aligning safety policies across the services that provide its models and updating its Usage Policies, effective November 20, 2026, to make the requirements clearer for developers.
- The company says it worked with customers and partners over recent months to set consistent safeguards whether people reach its models directly through OpenAI, through partners that host them, or through reseller partners.
- Under the policies, partners providing access to OpenAI models must maintain safeguards consistent with OpenAI's own, investigate potential violations, and take action when concerns are identified, including coordinating on enforcement so restrictions hold across services.
- The updated Usage Policies require services offering direct access to OpenAI models, such as model routers or AI gateways, to get OpenAI's review and approval before offering that access.
- OpenAI says it evaluates a model's capabilities and risks before deployment, trains it to follow safety requirements, and tests whether safeguards hold up when people try to bypass them, then monitors for abuse after deployment.
- The company is also working with customers on SafetyIDs, which let it distinguish between individual end users in an investigation without learning those users' identities, and says it bases enforcement on documented conduct and risk and revisits restrictions when conduct, commitments or safeguards change.
