Anthropic, the artificial intelligence start-up behind the Claude chatbot, has notified users of a significant update to its privacy policy. Starting July 8, the company will require certain individuals to prove their age or identity to access specific features.
In an email sent to users on June 22, the company stated the measure is part of its efforts to ensure the safety and security of its services. “As part of the steps we take to ensure the safety and security of our services, we may ask you to verify your age or identity,” the notification read, directing users to the updated policy.
Limited Rollout for Fraud Prevention
While the need for age verification is clear—to confirm a user is over 18—the requirement for identity checks is more nuanced. Anthropic elaborated on a recently updated support page, explaining the verification will appear when users access certain functionalities as part of regular platform integrity controls or other security and compliance measures.
An Anthropic engineer further clarified on social media platform X that the process is not a blanket policy. The verification applies only to a small subset of users flagged for potentially fraudulent activity, offering them a path to compliance rather than facing an outright ban. The engineer also confirmed the deployment is entirely unrelated to the recent forced withdrawal of the Mythos and Fable models, which occurred after a White House order required Anthropic to block access for foreign nationals—a technical segregation the start-up found impossible to implement.
How the Verification Process Works
To prove their identity or age, users will be required to submit a photo of an official government-issued ID, such as a passport, national identity card, or driver’s license, alongside a selfie in photo or video format. The company claims the identity verification process typically takes less than five minutes.
Anthropic will collect this documentation, as well as facial geometry patterns—which may be classified as biometric data in certain jurisdictions—and the outcome of the check. In a bid to assuage privacy concerns, the firm explicitly stated that this sensitive data will not be used to train its artificial intelligence models.
