Claude Identity Verification: Who Has to Upload a Passport (and Who Doesn't) Before July 8
On July 8, 2026, Claude identity verification becomes mandatory for some users. Anthropic started testing it back in mid-April, quietly, on what the company called "a few use cases." Then a June privacy-policy update put a hard date on it. Cue the panic headlines about Claude wanting your passport and your face. Here's the calmer truth: most people reading this will never see a verification prompt, and if your work runs on the Anthropic API or Claude Code's API path, it's almost certainly untouched. Let's sort out who's actually affected.
Key takeaways:
- Mandatory from July 8, 2026; Anthropic began testing it around April 14-15, 2026.
- Applies to Claude.ai web and Claude Code consumer surfaces, not the direct Anthropic API or Developer Platform.
- Verification means a government photo ID plus a live selfie, handled by Persona; most users are never prompted.
- Free, Pro, and Max accounts are in scope; Team, Enterprise, and Developer Platform run on separate terms.
What Is Claude Identity Verification? (And Why Now)
Claude identity verification is Anthropic's requirement, mandatory from July 8, 2026, that some Claude.ai and Claude Code users confirm their identity with a government-issued photo ID and a live selfie through Persona. It targets abuse, age compliance, and policy enforcement. Direct API and Developer Platform access is not affected.
So why now? Anthropic frames this as a platform-integrity move, not a blanket data grab. The verification is triggered by specific signals: requests tied to "certain capabilities," accounts that look like they're operating from an unsupported region, age-verification needs, or fraud-style patterns. According to the Anthropic Help Center, this is rolling out "for a few use cases," not for everyone at once.
The timeline is the part the news cycle blurred. Testing started in mid-April 2026 (reporting puts it around April 14-15). The June privacy-policy update formalized it and locked in July 8, 2026 as the date the requirement becomes enforceable. This lands during the same stretch as another recent change to Claude accounts, so if your Claude experience has felt different lately, you're not imagining it.
Most Claude users will never see a verification prompt. It's targeted, not blanket. That single sentence is what the passport-panic headlines left out.
Not the same as a "Verified Organization" gate. If you searched this expecting OpenAI's developer-side org verification, this isn't that. Anthropic does not require ID to unlock API models or tiers. This is a consumer account check, not a developer-platform unlock.
Who Actually Has to Verify? (Plan-by-Plan)
Free, Pro, and Max accounts are in scope for Claude identity verification because they're consumer surfaces. Team, Enterprise, and Developer Platform customers are covered by separate commercial terms and sit outside this consumer gate. Being "in scope" does not mean you'll definitely be asked. The prompt is targeted.
Here's the clean breakdown:
| Plan / Surface | In scope? | Notes |
|---|---|---|
| Claude Free | Yes | Consumer plan, can be prompted if signals trigger it |
| Claude Pro | Yes | Consumer plan, targeted prompts only |
| Claude Max | Yes | Consumer plan, targeted prompts only |
| Claude Team | No | Separate organizational terms |
| Claude Enterprise | No | Separate organizational terms |
| Developer Platform / API | No | Commercial terms, excluded from this gate |
The distinction that trips people up: in scope is not the same as prompted. A normal Pro user in a supported region, behaving normally, has low odds of ever seeing the screen. Anthropic built this to catch suspicious or policy-relevant activity, not to fingerprint every paying customer. Free, Pro, and Max are in scope; Team, Enterprise, and Developer Platform run on separate terms.
Does It Affect the API or Claude Code? (Developer Decision Table)
No, the direct Anthropic API does not require identity verification. Claude.ai web and Claude Code's consumer surface can prompt for it, but API access authenticates with API keys, not ID documents. Your CI pipelines, production calls, and gateway-routed API traffic are unaffected by Claude identity verification.
This is the part most coverage got half-right, so let's be precise:
| Surface | Verification required? | What it means for you |
|---|---|---|
| Claude.ai web | Yes | Can prompt if signals trigger it |
| Claude Code (consumer login) | Yes | Can surface mid-session on a consumer account |
| Direct Anthropic API (API key) | No | Auth is the API key, not your ID |
| API via gateway / multi-provider | No | Routed traffic stays unaffected |
| Bedrock / Vertex / Foundry | No | Separate provider terms entirely |
The nuance worth internalizing: Claude Code can authenticate two ways. If you log in with a consumer Claude account, you're on a consumer surface and the prompt can appear. If you point your Claude Code workflow at an API key, you're on the developer path, which this gate doesn't touch. That's the whole trick.
If your workflow runs on the direct Anthropic API, this verification gate does not touch it. Production code that calls Claude through API keys keeps running exactly as it did, which matters when you think about how developers use the API at scale. Worried about what else is shifting? Here's what's coming to Claude Code on the feature side. If your team is sorting out how an AI vendor policy change affects your pipeline, that's the kind of thing we help with: get a free consultation.
What You'll Need: Accepted vs Rejected IDs
If you're prompted, Claude identity verification needs a clear photo of a physical government-issued ID plus a live selfie through Persona. Passports, driver's licenses, state or provincial ID cards, and national identity cards work. Photocopies, screenshots, digital IDs, and non-government IDs do not. Anthropic says the whole thing typically takes under five minutes.
| Accepted | Rejected |
|---|---|
| Passport | Photocopies or scans |
| Driver's license / state ID | Screenshots or photos of a photo |
| National identity card | Digital or mobile IDs |
| Physical, legible, undamaged ID with photo | Student IDs, employee badges, bank cards |
| Temporary paper IDs |
A few practical notes. The ID has to be the physical card, clearly legible and undamaged. A photo of your screen showing a digital license won't pass, even though it feels like it should. The live selfie exists so Persona can match a real face to the document, which is why a still photo of yourself won't work either. If a check fails, you're typically allowed to retry with better lighting or a different accepted ID. "Under five minutes" is Anthropic's claim, not a promise. Plan for closer to ten if your camera or lighting fights you.
How the Verification Actually Works (Based on Anthropic's Documented Flow)
Based on Anthropic's documented Persona flow and widely reported user runs, the sequence is short: you're prompted in-product, you upload a government ID, you take a live selfie, then Persona processes the match. We did not personally trigger a live ID-and-selfie check on a techsy account, so what follows is the documented flow plus reported experience, clearly labeled as such, not a fabricated screen-by-screen of our own.
Here's the genuine step order, drawn from the help center and reported runs:
- The prompt appears. It surfaces inside Claude.ai or Claude Code, not as a separate app. On Claude Code's consumer login, it can interrupt mid-session rather than only at sign-in, which is the gotcha worth knowing.
- ID upload. You photograph or upload a physical government-issued ID. Persona checks it's legible, undamaged, and a real document, not a photo of a photo.
- Live selfie. You take a live selfie so Persona can match your face to the ID photo. A static image won't pass; it wants a live capture.
- Processing. Persona runs the match and returns a result, usually within a minute or two.
On timing: Anthropic says under five minutes, and clean runs reportedly land there. The realistic failure points are lighting, a glare on the ID, or a webcam that's lower quality than a phone camera. Reported runs suggest the phone capture path is smoother than a desktop webcam. If your first attempt bounces, the fix is almost always better light and a steadier shot, not a different ID.
To be straight with you: nobody, including us, should be presenting invented screenshots of this flow as proof. The honest version, documented steps plus real timing claims, tells you everything you need without faking a screen.
What Happens to Your Passport and Selfie? (Persona + Data Handling)
Anthropic uses Persona, a third-party verification subprocessor, to handle the ID and selfie. Anthropic says the data is "encrypted in transit and at rest" and that it is "not using your identity data to train our models." The images are collected and stored by Persona, not on Anthropic's own servers, which is the detail that changes the privacy math.
Persona (Persona Identities) is the company doing the actual document and face matching. Anthropic's stated reason for picking them is "the strength of their technology, privacy controls, and security safeguards," per the help center. The official privacy phrasing is worth attributing carefully, and it's spelled out in Anthropic's privacy policy: encrypted in transit and at rest, not used for model training, processed for security and compliance.
Now the honest part. Anthropic says your ID and selfie aren't used to train its models, and that's a meaningful commitment. But you are still handing biometric data to a third party, and "stored by Persona, not Anthropic" means your trust is split across two companies instead of one. That's not a reason to panic. It is a reason to read who Persona is before you upload anything, especially if you're in a region with strict biometric-data rules.
Should You Actually Be Worried? An Honest Take
Probably not, if you're a normal user in a supported region. Claude identity verification is targeted at suspicious activity, fraud signals, age checks, and "certain capabilities," not at every account. Most affected users will never see a prompt. The real concern isn't surveillance of everyone; it's the biometric trade-off for the minority who do get checked.
Let me give you the calm verdict the news cycle skipped. If you're a paying Pro user, working normally, from a country Anthropic supports, the odds you'll ever be prompted are low. The system is built to catch edges, not to census the middle.
There's one edge case worth owning honestly. Verifying your identity is not a free pass. The help center is clear that accounts can still be banned for things like repeated Usage Policy violations, creating an account from an unsupported location, or under-18 usage. Verification confirms who you are; it doesn't retroactively approve how the account was used. So if you verify and still get restricted, that's not a bug, it's the policy working as written.
The genuine privacy concern stands: handing a passport and a live selfie to any vendor is a real decision, not a formality. But the panic framing, that Claude is mass-collecting faces, doesn't match what the documentation describes. If you want the wider context on where Claude is heading, our broader Claude ecosystem guide covers the product side.
Frequently Asked Questions
When does Claude identity verification start?
It becomes mandatory on July 8, 2026. Anthropic began testing it around April 14-15, 2026, on a limited set of use cases, then a June privacy-policy update set the firm deadline. Between now and July 8, prompts are rolling out gradually rather than all at once, so timing varies by account.
Does Claude identity verification affect the API?
No. The direct Anthropic API and Developer Platform are excluded. API access authenticates with API keys, not identity documents, so your production calls, CI pipelines, and gateway-routed traffic keep working unchanged. This consumer verification gate covers Claude.ai web and Claude Code's consumer login, not your developer integrations.
Does Claude Code require identity verification?
It can, but only on the consumer login path. If you sign into Claude Code with a consumer Claude account, the prompt can surface mid-session. If you authenticate Claude Code with an API key instead, you're on the developer path, which this verification gate does not touch.
What ID does Claude accept?
Government-issued physical photo IDs: a passport, a driver's license or state/provincial ID card, or a national identity card. The document must be legible, undamaged, and include a photo. Photocopies, screenshots, digital or mobile IDs, student IDs, employee badges, and temporary paper IDs are all rejected.
Is Claude identity verification mandatory for Pro users?
Pro accounts are in scope, but in scope is not the same as prompted. The check is targeted at suspicious activity, fraud signals, age verification, and certain capabilities. A normal Pro user in a supported region, behaving normally, has low odds of ever being asked to verify.
Can I use Claude without verifying my identity?
For most people, yes, because most accounts are never prompted. But if you do get a verification prompt and decline it, expect that capability or account access to be restricted until you complete the check. There's no documented permanent way to skip a prompt once it's triggered on a consumer surface.
Who is Persona, and is my passport safe?
Persona (Persona Identities) is Anthropic's third-party verification subprocessor. Anthropic says the data is encrypted in transit and at rest and isn't used to train its models, and the images are stored by Persona rather than Anthropic. The real trade-off is that you're trusting Persona with biometric data, so read their handling terms first.
Why is Anthropic asking for my ID?
Anthropic frames it as platform integrity: abuse prevention, age compliance, fraud signals, unsupported-region checks, and gating "certain capabilities." It's positioned as a targeted security measure rather than blanket data collection, which is why most users never trigger a prompt in the first place.
Is this the same as OpenAI's "Verified Organization" developer gate?
No. OpenAI's org verification unlocks developer models and tiers on the API side. Anthropic's Claude identity verification is a consumer account check on Claude.ai and Claude Code's consumer login. Anthropic does not require ID to unlock API models or tiers, so the two systems serve different purposes.
What happens if my account is banned after verifying?
Verification confirms who you are; it doesn't approve how the account was used. The help center lists ban reasons including repeated Usage Policy violations, account creation from an unsupported location, Terms of Service violations, and under-18 usage. So you can verify successfully and still face restrictions if those apply.