To use Claude's top models, scan a passport and a live selfie
A government ID, a face camera, under five minutes. Anthropic's verification flow routes your encrypted photos to a third party called Persona, not to Anthropic itself. The twist surfaced in the thread: this support page has quietly existed since April.
Photocopies, wallets and work badges all get rejected; only a passport, license or national ID plus a live selfie clear the gate. The stated goals are to prevent abuse, enforce policy and meet legal obligations, and the check can fire mid-session.
Anthropic never holds the images: Persona keeps them encrypted and walled off from training. What looked like a fresh crackdown turned out to be an old page resurfaced, and the suspicion was less about privacy than who is leaning on Anthropic to build the gate.
A passport, a selfie, and a five-minute timer
The Claude support article reads like routine onboarding copy, which is part of why it landed so hard. Anthropic verifies identity, it says, to "prevent abuse, enforce usage policies, and comply with legal obligations." The check can fire when you reach for certain capabilities, or during an ordinary integrity and safety review with no warning at all.
What the flow actually demands
The mechanics are specific. You hand over a government photo ID — passport, driver's license, or national ID card — and then a live selfie, a process Anthropic says usually finishes in under five minutes. Photocopies are out. Digital IDs are out. Badges are out. The work is done not by Anthropic but by a vendor, Persona Identities, which holds the encrypted images, is contractually limited to the verification task, and never feeds them into model training.
Fail the check and you get retries plus a support form. Accounts can still be suspended for ToS violations, for sitting in an unsupported location, or for being under 18, each with an appeals path. On paper it is careful. In the thread the carefulness was beside the point.
The page nobody announced
The loudest correction was chronological. The framing of a sudden new mandate collapsed once readers checked the timestamp.
"Not quite - this page has been there and saying this since April, so ~3 months ago." — the developer behind donohoe.dev
With the "breaking news" reading gone, attention moved to motive. One reader connected the dots to export controls and a stalled model rather than to safety theater.
"I suspect this new Anthropic requirement is coming from their ongoing negotiations with the government to re-enable Fable." — a commenter who notes OpenAI already had an ID check in place
Those who had run OpenAI's comparable gate came with a warning about the failure mode, not the principle.
"What is especially bad is that if you fail the verification process, they won't let you retry – you are permanently locked out from the top models." — JimDabell
And the choice of vendor carried its own history.
"Persona as the provider is a bit surprising and problematic. Discord dropped them after user backlash." — a commenter who links the prior HN discussion of Persona
The gate was always there
The story is not that Anthropic suddenly demanded your ID; it is that the demand was sitting in a support article for three months and nobody noticed until someone resubmitted it. That reframes the anger. If the trigger is government pressure tied to export controls and a blocked model, then the verification page is a downstream artifact of negotiations users never see, and the permanent-lockout risk that OpenAI users describe is the part to read closely before you raise a camera to your face.
Get breakdowns like this every day
The top Hacker News stories, unpacked — by email, Telegram, or RSS.
When both résumé and interview are AI, hiring selects for performers
Google and McKinsey are flying candidates back into rooms, eating the travel cost on purpose. The reason, two startup founders argue, is that the cheap early filters now reward whoever games the process best — and an AI detector says the essay making that case was itself written by AI.