Anthropic has begun asking some Claude users to verify their identity with a government photo ID and a live selfie, handled by KYC vendor Persona. Its updated privacy policy extends the checks to the Free, Pro and Max tiers from 8 July.
The support page is blunt about what Persona collects: the ID document and everything printed on it, a face photo or video, and "facial geometry templates" that Anthropic concedes "may be considered biometric data in some jurisdictions." Anthropic is the data controller; Persona only processes on its instructions.
The why-now is agentic. A conversational chatbot does not need to know who you are, but an agent booking flights or moving documents on your behalf raises a real authorisation question, and it pushes general-purpose AI providers toward the compliance posture of a bank. One awkward note for an identity vendor: researchers found a Persona government-dashboard codebase sitting on a public endpoint in February.