Anthropic says Claude may want to see your ID

Anthropic whitethorn inquire Claude users to verify their property and individuality by uploading their government-issued documents, according to a caller mentation of the company’s privateness policy. 

The AI elephantine says the determination was to let users to entreaty having their relationship flagged for perchance fraudulent enactment alternatively than outright banning them, but comes astatine a clip arsenic Anthropic seeks to placate the Trump medication amid an ongoing standoff implicit who gets entree to the company’s AI tools.

According to a caller conception successful its latest privateness policy published earlier successful June and acceptable to instrumentality effect connected July 8, Anthropic says it volition inquire for a idiosyncratic to beryllium their property oregon individuality “in definite circumstances,” without providing circumstantial examples. 

While Anthropic has agelong required users to beryllium over-18 to usage Claude, the institution earlier this twelvemonth introduced age verification checks to comply with assorted states and countries that necessitate them. Identity checks were also announced, but weren’t reflected successful the company’s privateness argumentation until much recently.

When triggered, the argumentation would necessitate those users to upload a photograph scan of a government-issued passport oregon driver’s license. Anthropic says it volition besides cod a person’s selfie photograph oregon video and the person’s digitized mentation arsenic a look geometry template (which immoderate states, similar Illinois, considers legally protected biometric data). Anthropic says it volition besides support a grounds of the verification result, specified arsenic whether the idiosyncratic has reached a definite age.

When reached by TechCrunch, Anthropic spokesperson Michael Aciman shared a nexus to an X post from Anthropic’s Thariq Shihpar saying that the alteration applies lone to a “small subset of users” whose accounts are flagged but not outright banned. (Anthropic’s spokesperson would not accidental however galore users is simply a subset, but the institution is thought to person tens of millions of users monthly.) 

“[Anthropic’s identity verification policy] was updated connected June 17 arsenic an update to the appeals process,” said Shihpar successful the post. “It’s unrelated to the Fable oregon Mythos rollout.”

Anthropic said it is allowed to necessitate users to upload a transcript of their IDs for a fig of reasons, specified arsenic for requiring users to verify themselves for creating and administering their Claude account, and enforcing its presumption of service, specified arsenic to forestall and analyse fraud, abuse, and violations of its terms, including unlawful oregon transgression conduct, and to analyse and resoluteness information issues.

The determination to support person tabs of who is utilizing Anthropic’s AI tools may beryllium one way for the institution to comply with a assortment of ongoing ineligible challenges, regulatory changes, and inbound pressures from the Trump administration.

The tech elephantine remains mostly astatine an impasse with the White House, much than a week aft Trump officials efficaciously forced Anthropic to propulsion its latest cybersecurity models over allegations that an evident jailbreak could interruption the models’ guardrails. Other reports person pointed to personality clashes betwixt the institution and the Trump administration arsenic the greater culprit of the breakdown successful relations.

This latest clash comes months aft the Department of Defense designated Anthropic a “supply concatenation risk”, seemingly successful retaliation for not allowing the authorities to usage its exertion for wide home surveillance oregon powering afloat autonomous weapons.

Anthropic said it uses the San Francisco-based institution Persona arsenic its identity checking provider, and that users whitethorn “see a verification punctual erstwhile accessing definite capabilities, arsenic portion of our regular level integrity checks, oregon different information and compliance measures.”

Anthropic said that it decides for however agelong Persona retains its users’ individuality documents, but Anthropic’s spokesperson did not instantly accidental erstwhile the information was deleted. (For context, Roblox, different lawsuit of Persona, says users’ images are deleted “immediately” aft they are processed, limiting the accidental of the accusation aboriginal being leaked oregon stolen.)

Persona tin inactive look U.S. authorities demands for users’ accusation that it stores connected its servers.

Persona is backed by Founder’s Fund, an concern steadfast founded by Trump backer Peter Thiel, who besides invests successful Anthropic. The individuality checking steadfast has faced disapproval from users for its links to Thiel. Earlier this year, Discord chose Persona for its property verification checks past rapidly reneged following idiosyncratic backlash for choosing Persona.