You Switched to Claude Over Surveillance Fears. Now It Wants Your Passport

Anthropic quietly revealed identification verification necessities for Claude this week, asking sure customers at hand over a government-issued photograph ID and a reside selfie. One thing its rivals don’t require.

“We’re rolling out identification verification for a number of use instances, and also you would possibly see a verification immediate when accessing sure capabilities, as a part of our routine platform integrity checks, or different security and compliance measures,” Anthropic mentioned. “We solely use your verification knowledge to substantiate who you might be and never for some other functions.”

Tens of millions of customers fled OpenAI for Anthropic in February after OpenAI signed a deal to deploy AI on Pentagon categorised networks—a contract Anthropic turned down over considerations about mass surveillance and autonomous weapons. Every day signups broke data, and free customers had been up 60% since January, Anthropic mentioned on the time. The privacy-conscious crowd had discovered its residence.

That crowd, it appears, might now have some paperwork to organize if it needs to proceed utilizing Claude. The reactions to date have been fairly damaging, declaring that it’s a deliberate determination and never a regulation or a compulsory order imposed by a authorities on Anthropic as a service supplier.

In accordance with the assistance heart web page, which went reside on April 14, Anthropic chosen Persona Identities as its verification companion—the identical KYC infrastructure used throughout monetary providers—and requires a bodily, undamaged passport, driver’s license, or nationwide identification card. Photocopies, cell IDs, and pupil credentials do not rely. A reside selfie might also be required.

The coverage is not common but. Verification will set off when accessing “sure capabilities,” throughout “routine platform integrity checks,” or as a part of security and compliance measures. Anthropic hasn’t mentioned publicly which options are gated, or what consumer habits would possibly immediate a verify. The corporate didn’t instantly reply to Decrypt’s request for extra particulars.

On knowledge dealing with, Anthropic attracts a cautious line: your ID and selfie go to Persona’s servers, not Anthropic’s personal programs. The corporate says it’s the knowledge controller setting the phrases, and that Persona can use the data to confirm identification and enhance fraud detection. The info is encrypted in transit and at relaxation, excluded from mannequin coaching, and will not be shared with third events for advertising and marketing, one thing Anthropic has been cautious to vow since its earliest business insurance policies.



Cautious guarantees, although, have a historical past of assembly careless infrastructure. An October 2025 breach at Discord uncovered roughly 70,000 authorities IDs customers had submitted for age verification. Persona is a severe participant on this house, however third-party custody of presidency paperwork has demonstrated repeatedly that no third social gathering is immune.

Tighter identification controls additionally match a sample Anthropic has been constructing towards. In December, the corporate introduced classifiers to detect customers who self-identify as minors. A number of grownup customers had their accounts suspended anyway, reporting that total challenge histories had been wiped whereas they tried to enchantment incorrect flags.

Accounts registered from areas Anthropic does not formally serve are additionally topic to bans—a element that lands hardest on Chinese language customers accessing Claude by way of intermediaries, since a reside selfie matched in opposition to a bodily authorities doc is tough to pretend your approach by way of.