Claude Wants Your Face and Your ID Starting July 8 — Read This First

Imagine you open the AI tool you've been using every week for months—maybe you use it for work, for research, for drafting emails—and instead of the usual chat box, there's a new screen. It's asking for your government-issued ID. And a selfie. And something called "facial geometry data" (which is a technical way of saying it wants to map the unique shape of your face). You blink. You re-read it. And you think: Is this real?

It's real. And it's coming faster than most people realize.

This Is Not a Normal Verification Step

When a website asks you to verify your email, you click a link. Done. When they ask for a phone number, you type in a code. Annoying, but low-stakes. What Anthropic is now asking for — government ID plus a live selfie plus a facial geometry scan — is a completely different category of information. It belongs in the same conversation as what your bank collects when you open an account, or what a federal agency holds on file.

Here's why the difference matters so much. According to academic research on biometric identity risks, biometric data — meaning your face, fingerprints, voice patterns, the physical stuff that is uniquely you — cannot be changed if it's stolen. Your password gets leaked? You change it in five minutes. Your facial geometry data gets exposed in a breach? You cannot get a new face. That asymmetry is what makes this kind of data so sensitive, and why most people have only ever shared it with governments and banks — not chatbot companies. This article is part of a series — start with Your Face Is The Ticket What Happens When The Computer Says .

The collection that TechCrunch first reported on June 22, 2026 isn't even happening through Anthropic directly. The company uses a third-party verification vendor called Persona Identities to run the checks. Your ID and facial scan go to Persona's servers — not Anthropic's. According to TechTimes, Persona can retain that data for up to three years, and its system is capable of running 269 distinct verification checks — including screening users against terrorism watchlists and flagging accounts to FinCEN, which is the U.S. Treasury's financial crimes enforcement network. That's a significant amount of power sitting behind a screen that just asked you for a selfie.

Why Is This Happening Now?

The honest answer is: because the U.S. government forced the issue, and Anthropic had no backup plan.

On June 12, 2026, a federal export control directive required Anthropic to block foreign nationals from accessing its most powerful AI models. The problem? Anthropic had no way to verify who was American and who wasn't. So the company did what companies do in a crisis — it suspended access to those advanced models globally, then built an ID verification system retroactively to restore access for people who could prove their nationality. As CIO reported, verifying U.S. identity through ID checks may offer a path for affected users to regain access to capabilities that were shut off — but it requires handing over exactly the kind of sensitive data most people have never given to a tech company before.

This wasn't a slow, thoughtful rollout. The quiet pilot started on April 14, 2026, covering what Anthropic described as "a few use cases." By June, it became formalized policy across the platform. As AI Agents First documented, the timeline went from quiet experiment to company-wide policy in under three months — without much public notice to users who might want to make an informed decision before July 8 arrived. Previously in this series: Your Dmv Photo Is Now A Biometric Profile And Nobody Asked Y.

"Anthropic can legally demand government IDs, live selfies, and facial geometry scans from consumer Claude users — the first time a major U.S. frontier AI lab has codified this level of biometric collection in consumer terms." — The Next Web, analysis of Anthropic's updated privacy policy

That phrase — "legally demand" — deserves a second look. Anthropic isn't doing anything illegal. But "legal" and "something you should hand over without thinking" are two very different things. Illinois has a law called the Biometric Information Privacy Act (BIPA for short — basically a state rule that says companies must get real consent before collecting face or fingerprint data and can be fined $1,000 to $5,000 per violation if they don't). Facebook got hit with a $650 million class-action settlement under that exact law for collecting facial data without proper consent. The legal teeth exist. Whether they bite in your situation depends on where you live and how the consent screen was worded.

The Question Nobody Is Asking Out Loud

Here's where it gets interesting. The industry argument — and it's not a stupid argument — is that this kind of identity verification actually protects you. When deepfakes (AI-generated fake videos and images) can now convincingly impersonate almost anyone, and when AI tools are powerful enough to be genuinely dangerous in the wrong hands, proving you're a real person with a real identity has a legitimate purpose. Banks have done this for years. Airports do it at the gate. The question isn't whether identity verification is bad. It's whether a consumer AI chatbot is the right place for the same level of scrutiny that the TSA applies at security checkpoints.

And look — nobody is saying Anthropic is behaving villainously here. The export control problem is real. The deepfake problem is real. Other major AI companies are moving toward similar requirements for their most powerful access tiers; Anthropic is just the first to put it directly in consumer-facing policy language this explicitly. That matters, because explicit policy means explicit data collection — and that's worth paying attention to regardless of which company is doing it.

The real warning isn't about Anthropic specifically. It's about what happens when you treat this as just another "verify your account" pop-up. Because mentally, most of us are conditioned to click through those without reading them. We've been trained by years of email confirmations and two-factor codes to see verification as a minor speed bump. This is not a speed bump. A government ID plus a live face scan plus a facial geometry map is a fundamentally different kind of handoff — and the fact that it now arrives in the same interface as a cookie consent banner does not make it the same thing. Up next: Digital Id Wallet Biometric Recovery Vulnerability.

If you've ever wondered whether a photo or a profile is really who it claims to be — if you've ever second-guessed a video call or a message from someone you thought you knew — that instinct is exactly why this kind of verification technology exists. It's trying to solve a real problem. The issue is that most of us have never been told what we're actually signing up for when we say yes to it.

One practical thing you can do right now, before any verification screen appears: look up the privacy policy of any AI tool you use regularly and search for the word "biometric." If it's there, find out who the third-party vendor is, look up that vendor's data retention terms, and then decide if the tool is worth the tradeoff. That's not paranoia. That's just reading the fine print before you sign — which, in this case, involves your face.

The Facebook BIPA settlement took years of litigation and cost $650 million — and Facebook still collected the data first, asked questions later. Anthropic's policy is at least upfront about what it wants. But here's the thing worth sitting with: the company built the ID verification system after it needed it, in under three months, in response to a regulatory crisis it didn't anticipate. That's not a reason to panic. It is, however, a pretty good reason to ask exactly how much thought went into what happens to your facial geometry data in year three.