Claude Will Soon Need to See Your ID

Welcome back to SavvyMonk, your one-stop for AI and tech news that actually matters.

Anthropic spent 2026 telling Washington it would not turn Claude into a surveillance tool. On 8 July it starts asking a slice of its own users to hand over a government ID and a live scan of their face.

Let's get into it.

Most AI tools ask you to trade your data for intelligence. Norton Neo doesn't. It's the first safe AI-native browser built by Norton, and it gives you powerful built-in AI without handing your privacy over to get it. Search, summarize, and write with AI built directly into your browser. Your data stays yours. Your context stays private.

Built-in VPN, anti-fingerprinting, and ad blocking come standard. No add-ons. No setup. No compromises.

Fast. Safe. Intelligent. That's Neo.

Download Norton Neo

Anthropic has quietly added a new category to its privacy policy called Verification Data, and the section takes effect on 8 July 2026. When the company decides a check is warranted, it can ask a user to upload an image of a government-issued passport or driver's licence along with every detail printed on it, including name, date of birth and document number.

It also collects a photo or video of the person's face and a digitised version of that face called a facial geometry template, which Anthropic itself concedes may count as biometric data in places such as Illinois. The policy says these checks happen in certain circumstances and offers no concrete examples, which is part of why the change has unsettled people who otherwise pay little attention to terms of service updates.

Anthropic has required users to be over eighteen for a long time, and earlier in 2026 it introduced age-assurance checks to satisfy a growing list of states and countries that demand them. Identity verification is the newer and far more sensitive layer.

How the Check Actually Works

The mechanics matter here. Anthropic is not letting people upload an old scan or a screenshot of a licence, and the support documentation rules those out explicitly.

The process runs as a live capture through Persona Identities, a San Francisco company that handles know-your-customer work for platforms ranging from OpenAI to Roblox. Anthropic describes itself as the data controller, meaning it sets the rules, while Persona acts as the processor that executes them. Your ID and your selfie sit on Persona's servers rather than inside Anthropic's own systems, and Anthropic says it can pull verification records when it needs to review an appeal.

The company also says it decides how long Persona keeps the documents, yet when a reporter asked for the actual retention window, Anthropic would not name one. For comparison, Roblox tells its users that Persona deletes their images almost immediately after processing, which limits the time a leak or a subpoena could expose them.

The Part That Worries People

The choice of vendor is where the story turns. Persona is backed by Founders Fund, the venture firm co-founded by Peter Thiel, who also happens to be an investor in Anthropic. Founders Fund led both Persona's Series C and Series D rounds, worth one hundred and fifty million dollars and two hundred million dollars respectively.

In February 2026 security researchers found roughly two thousand four hundred of Persona's front-end files sitting on a publicly accessible, government-authorised endpoint. The exposed code showed the platform can run two hundred and sixty-nine separate verification checks, screen faces against watchlists, scan adverse media across fourteen categories and file suspicious activity reports with federal agencies.

Persona chief executive Rick Song denied any surveillance role and argued the exposure was harmless front-end code rather than a breach. Discord, which had been trialling Persona for age checks, dropped the vendor within weeks of the backlash.

What Anthropic Says It Is Really Doing

Anthropic frames all of this far more narrowly than the headlines suggest. Spokesperson Michael Aciman pointed reporters to a public statement from Anthropic's Thariq Shihipar, who said the update touches only a small subset of users whose accounts have been flagged but not outright banned, and that it functions as a fix to the appeals process rather than a broad new gate.

Shihipar added that the change is unrelated to the Fable or Mythos model rollout, heading off speculation that identity checks were tied to the company's most capable systems.

Anthropic also runs age-only checks through a different vendor called Yoti, which returns a simple pass or fail and never passes the underlying ID image back to Anthropic. So the worst-case reading, where every one of Anthropic's tens of millions of monthly users suddenly faces a face scan, is not what the policy actually describes.

The Standoff Behind the Change

It is hard to read this move outside the fight Anthropic is having with Washington. The company is more than a week into an impasse with the White House after Trump officials effectively forced it to pull its latest cybersecurity models over claims that a jailbreak could defeat their guardrails.

Months earlier the Department of Defense had labelled Anthropic a supply chain risk, apparently as payback for refusing to let the government use Claude for mass domestic surveillance or autonomous weapons. That history is exactly why the Persona partnership lands so awkwardly. A company that built part of its public identity on declining to power surveillance is now asking ordinary users to verify themselves through a firm bankrolled by the same investor behind Palantir.

Persona can still be compelled to hand stored data to the United States government, which is the quiet risk sitting underneath the convenience.

What Is Still Unanswered

Several gaps remain even after the policy goes live on 8 July 2026. Anthropic has not published the list of behaviours that trigger a check, so users cannot know in advance whether a flagged login, a refused request or a subscription upgrade is what summons the prompt.

The retention period for ID images sits undefined, which makes it impossible to judge how long a copy of your passport lingers on a third party's servers. And because facial geometry templates carry legal weight in certain states, anyone in those jurisdictions is effectively handing over protected biometric data to clear an appeal.

None of this makes the system uniquely alarming compared with what Reddit, Discord or OpenAI already ask for, but it does mean the burden of trust falls on a vendor whose recent months have been defined by exactly the kind of exposure people fear.

The Bottom Line

Anthropic is not forcing a face scan on everyone, and the narrow appeals-process framing is probably accurate. But choosing Persona, of all vendors, to verify the users of a company that made its name refusing to build surveillance tools is a strange look, and the unanswered retention question is the part worth watching. If you never get flagged, nothing changes for you. If you do, the price of getting your account back is now your ID and a map of your face.

Category: Privacy Review

"Read the privacy policy at [paste URL] and give me a plain-English summary of exactly what personal data this service collects, which third party processes it, how long they keep it, and what I cannot opt out of. Flag anything involving biometric data or government ID, and list the specific questions I should email the company to close the gaps."

Every platform you use is slowly converging on the same demand, which is proof that you are a real, adult, traceable person. The trade is convenience and access in exchange for a permanent record of your face held by someone you did not choose. Worth deciding now how much of that you are willing to give, and to whom. Hit reply, I read every response.

See you in the next one.

— Vivek

P.S. Know someone who reads the privacy policies so the rest of us do not have to? Forward this to the most privacy-obsessed person you know. They can subscribe at https://savvymonk.beehiiv.com/