Anthropic wants your ID before you chat with Claude 🤖

Cliquez ici pour lire en français

Over the past week or so, some Claude users opened an unusual email: Anthropic is warning them that its privacy policy is about to change, with a new data category called « Verification data. » Starting July 8, 2026, the company reserves the right to ask you to prove your age or identity to access its assistant — even if you’re just a free or paying subscriber.

What’s actually changing for your account 🧾

In practice, your relationship with Claude could stop being anonymous. Instead of an account tied to nothing more than an email address, you’d have a profile linked to a passport, driver’s license, or ID card — sometimes alongside a live selfie video. Anthropic frames the shift as a security and compliance measure, but it opens a new privacy battle for the generative AI era.

For now, the change only applies to consumer accounts: Claude Free, Pro, and Max — the plans you and I actually use through Anthropic’s regular interfaces. Team, Enterprise, and the developer platform fall under separate contractual terms and don’t (yet) mention systematic identity checks.

If you’re flagged, Anthropic could ask you to upload an official ID: passport, driver’s license, national ID, or state/province card — no photocopies, no screenshots, no temporary documents. In some cases, a live selfie taken through your webcam or phone will be required to match your face to the document, which means processing biometric data such as facial geometry templates.

This verification reportedly won’t apply to everyone the moment the policy takes effect on July 8: Anthropic only mentions « certain circumstances, » and early signals suggest developers and accounts flagged as higher-risk — or potentially used by minors — are among the first in line. If you refuse, the company hints that account access could be suspended or significantly limited, though it hasn’t spelled out exactly how that would play out.

Persona, the third party that sees your face 🕵️

Here’s the key detail: Anthropic isn’t running the verification itself. It’s outsourcing the job to Persona, a KYC (Know Your Customer) vendor that already handles identity checks for other digital platforms. In practice, that means Persona’s servers are the ones receiving the photo of your ID, your selfie, and the metadata that comes with it — Anthropic only sees the verification results through that vendor’s interface.

Officially, Persona encrypts data in transit and at rest, and Anthropic says these images won’t be sold, used for advertising, or fed into AI model training. Persona is reportedly contractually limited to using the data to provide and improve its verification services and to fight fraud.

That reassuring picture gets murkier once you look at the vendor’s track record. Back in February, researchers found a Persona test interface left openly accessible online, exposing more than 2,400 user-related files. The company said it was an isolated environment with no personal data involved, but the episode was enough to raise doubts about the operational reliability of a vendor now tasked with centralizing the passports and selfies of Claude’s users.

Anthropic and Washington: an already tense relationship 🇺🇸

This privacy policy update lands in an already tense climate between Anthropic and the Trump administration — even though, chronology in hand, the two stories aren’t actually connected. In early June, Washington ordered the startup to cut off access to its flagship models, Fable 5 and Mythos 5, for « any foreign national, whether inside or outside the United States, » citing national security. Unable to clearly distinguish American users from foreign ones beyond an email address or a credit card, Anthropic chose to simply suspend global access to both Fable 5 and Mythos 5.

You might read that as proof Anthropic is finally arming itself with the tools to comply with this kind of government order — to verify who you are, where you’re from, and potentially filter access based on your nationality. But the timeline says otherwise: Persona-based identity verification first started testing on April 14, 2026, and the announced expansion to Free, Pro, and Max accounts was published around June 8 — before Fable 5 and Mythos 5 even launched, let alone got blocked in mid-June. « Verification data » isn’t a response to that episode; the two stories are running in parallel, with no established causal link.

Still, both stories point to the same underlying tension: governments’ growing leverage over access to increasingly strategic AI systems, and the open question of how far a company should go in identifying its users in order to comply.

What this verification push says about our digital future 🔮

The story of Anthropic, Persona, and your ID papers isn’t just another entry in the endless saga of terms of service nobody reads. It crystallizes a defining tension of our moment: we want AI that’s powerful, reliable, and accountable — but we also want to keep control over who knows what about us.

By tying Claude to potential identity checks, Anthropic is trying to reassure regulators and governments — while risking the trust of part of its user base, especially the most tech-savvy and privacy-conscious users. What’s at stake here isn’t just one startup’s policy; it’s a fault line: between a future where AI access depends on the full transparency of our identities, and one that tries to reconcile power with anonymity, accountability with the right to stay private.

It’s now up to you to decide whether proving you’re clean to your own AI is a price worth paying to keep using Claude — or whether that’s the line you’re not willing to cross.