Anthropic and OpenAI Tighten AI Access Rules in November

Anthropic and OpenAI are introducing separate controls on where their AI services can be used and how paid access is shared, with changes taking effect in November for users and outside platforms.

Anthropic's updated rules take effect Thursday, Nov. 12. Announced Oct. 8, they prohibit Claude use from mainland China and other regions where the service is unsupported.

The company began rolling out identity checks in April. A check may require a government ID and a selfie submitted through Persona. Anthropic lists unauthorized resale of Claude access and use from restricted regions as risk signals that can prompt verification; failed checks can lead to account closure.

OpenAI announced Oct. 21 that third-party platforms providing direct access to GPT models must get approval in advance. The requirement starts Nov. 20.

OpenAI is also limiting use across multiple paid accounts. A payment card tied to too many subscriptions can prevent a new account from buying a plan, while a device used across too many paid accounts can trigger restrictions. The limits cover Go, Plus and Pro plans.

Eligible Pro subscribers can receive API credits equal to their subscription fee, capped at $500 per month. OpenAI prohibits sharing accounts and reselling the credits; violations can lead to account closure.