Anthropic is requiring government IDs and live selfie scans from some Claude users, with verification handled by Persona Identities. Certain consumer tiers could face a biometric checkpoint, while business and API tiers are exempt.
Your ID images and facial scan land on Persona’s servers, not Anthropic’s. Anthropic says it can access those records for appeals. The stated goal: prevent fraud, enforce usage rules, meet legal obligations. Anthropic is explicit that verification data won’t feed model training.
The rollout lands in a turbulent stretch. Earlier in June, Anthropic launched and then suspended its Mythos model after the U.S. government raised security concerns about how the company could use the technology. In May, Anthropic closed a $65 billion Series H at a $965 billion valuation, becoming the most valuable AI startup in the world.
Here’s what the consumer-versus-business split actually signals: Anthropic is drawing a line between the tier where fraud, underage use, and policy violations are hardest to catch, and the tier where customers have already signed contracts. The exemption for Team, Enterprise, and API plans isn’t a coincidence — it’s where commercial accountability already exists. If your product is built on the API, you won’t see this. If you’re shipping directly to consumers on Claude’s free or paid tiers, you’ve just inherited a new friction point in your onboarding flow. And if your product relies on consumer accounts rather than the API, now’s the time to reconsider that architecture.
Nathan Zakhary