May ask you to show your ID when restoring access.

On June 17th, 2026, Anthropic changed its privacy policy to permit Claude to request government-issued identification documents during the appeals process. This change takes effect on July 8th, 2026. In the event that your account has been flagged for suspicious activity or a violation of our terms of service, we may provide an opportunity for you to appeal. This process includes sending your passport or driver’s license, taking a selfie, and sharing a facial geometry scan. This is a modification to the appeals process and only affects a small fraction of users whose accounts are being reviewed.

Persona, a San Francisco-based identity verification provider that partners with various consumer services, will handle the verification of your identity documents. They decide how long to store your information and biometrics (which the new Anthropic privacy policy does not outline). As a US-based company, Persona is also subject to requests from US law enforcement. According to Anthropic, this verification process enables them to comply with state and national age verification laws, reduce fraud and abuse, and have a tool to investigate potential TOS violations when they would otherwise need to delete the account.

1Why this is bigger than it looks

Two years ago, asking an AI chatbot to collect biometrics and government IDs seemed outlandish. But this is where regulation is heading. With the EU AI Act, state-level online safety bills in the US, and numerous countries requiring age verification, platforms are being forced to know their user’s true identities — not just their username and email address. While this change is framed as only affecting a portion of users who go through the appeals process, it is also laying the groundwork. Once you have an ID verification flow set up for appeals, it’s relatively easy to reuse it for age gates on certain features, adhering to various regional regulations, or preventing coordinated abuse that is hard to do with anonymous accounts.

It’s also worth considering the timing of this announcement. Anthropic is publicly fighting with the US government about access to their strongest models while also facing regulatory scrutiny on several fronts. By building a system to identify who their users really are instead of just using their self-reported accounts, Anthropic has acquired tools they didn’t have before for compliance, risk mitigation, and enforcing their terms of service in cases where there are high stakes. Although these reasons aren’t explicitly stated in the privacy policy change, Anthropic now has the capability.