Claude identity verification launched quietly around April 14-15, 2026, requiring users to submit government-issued photo identification and live selfies to unlock certain advanced capabilities. Anthropic’s move reflects a broader industry shift toward gating AI features behind identity checks, but the rollout has exposed sharp divisions between users who see reasonable safety guardrails and those who view it as unnecessary surveillance.
Key Takeaways
- Claude identity verification requires passport, driver’s license, or national ID plus a real-time selfie via camera.
- Third-party Persona Identities handles verification; ID images stored encrypted off Anthropic’s systems.
- Process takes under 5 minutes but blocks access for users without accepted government photo ID.
- Anthropic states biometric data is not used to train Claude’s models.
- Chinese users face particular friction due to passport requirements and geopolitical data privacy concerns.
How Claude Identity Verification Works
The Claude identity verification process is straightforward on the surface but reveals complications beneath. Users prepare a valid government-issued photo ID—passport, driver’s license, or national ID card—and a device with a camera. They then capture a real-time live selfie while holding the ID, submit both via Persona Identities’ platform, and wait for verification, typically completed in under five minutes. Anthropic selected Persona based on its technical capabilities, privacy controls, and security safeguards.
The critical detail: Persona stores ID images and biometric data encrypted on its own systems, not Anthropic’s servers. Anthropic retains access only to verification records—a distinction that matters for users worried about centralized AI company data breaches. Still, this outsourcing to a third party introduces a new attack surface. Persona itself experienced a data exposure in April 2026, when Discord users’ verification data leaked and LinkedIn information was shared with up to 17 companies. The incident occurred independently of age verification issues, but the timing raises legitimate questions about whether Persona is the right custodian for biometric data at scale.
Which Claude Features Now Require Identity Verification
Anthropic has not publicly specified which advanced capabilities trigger the verification prompt. The company states that verification is prompted for certain advanced features, routine platform integrity checks, safety and compliance measures, or age restrictions—with accounts belonging to users under 18 facing suspension. This vagueness is intentional; Anthropic likely wants flexibility to gate different features for different user cohorts without announcing each decision publicly.
The lack of transparency fuels skepticism. Users cannot predict when they will hit the verification wall or which features sit behind it. This creates friction at unpredictable moments rather than at signup, a design choice that feels more like a compliance checkbox than a thoughtful user experience.
Privacy Claims and the Biometric Data Question
Anthropic explicitly states that Claude identity verification data is not used to train its models. That statement is important—and incomplete. Not using data for model training does not mean data is never used. Anthropic could theoretically use verification records for safety audits, abuse detection, or user behavior analysis without feeding biometric samples into model weights. The company’s narrow assurance leaves room for interpretation.
The deeper issue: requiring a live selfie for feature access creates a biometric record linked to account behavior. Even encrypted and held by a third party, that record represents a new form of user tracking. Credit cards, which minors can possess, are insufficient for age verification—yet a selfie, which reveals far more personal information, becomes the standard. This escalation deserves scrutiny. Users accepting the tradeoff should understand that they are trading biometric data for access to capabilities that worked without this requirement weeks ago.
Who Gets Locked Out
Claude identity verification creates a hard barrier for users without accepted government photo ID. Aadhaar cards, India’s national ID system, are not detected by Persona. Chinese users face a particularly acute problem: many lack passports, and those who do face geopolitical concerns about sharing biometric data with a US company. For new Chinese users trying Claude for the first time, the verification requirement acts as a de facto block on account creation if they cannot produce a passport.
Anthropic’s rollout is gradual and affects certain users globally, meaning some people will encounter the prompt while others do not. This uneven deployment raises fairness questions. Why should a user in one region face biometric verification while a user in another region does not? Anthropic has not addressed this disparity.
Does Identity Verification Actually Prevent Misuse
Skepticism about the security value of ID verification is warranted. Hacker News users noted that Claude’s AI refusal system and telemetry are more effective at preventing misuse than static ID scans. A determined bad actor can obtain a fake ID or use a borrowed passport. The friction imposed on legitimate users may exceed the friction imposed on adversaries. Persona support has also drawn criticism for poor responsiveness, with users reporting copy-paste responses that fail to resolve issues.
This is not to say identity verification has zero value. Age gating, compliance with regional regulations, and abuse account linking are genuine use cases. But framing the system as a security breakthrough is overselling it. It is a compliance and friction tool, not an impenetrable gate.
What This Means for the Broader AI Industry
Anthropic’s move signals a trend: as AI capabilities become more powerful and regulatory pressure increases, companies will experiment with gating features behind identity verification. OpenAI, Google, and others are watching. If Anthropic’s system succeeds in reducing regulatory complaints without tanking user growth, expect similar systems to proliferate. If it creates enough friction that users abandon Claude for competitors, the industry will reconsider.
The precedent matters. Today it is identity verification for advanced Claude features. Tomorrow it could be biometric verification for any AI access, or mandatory government registration to use certain models. Each step feels incremental and justified individually. Collectively, they reshape what it means to access AI in public.
Is Claude identity verification required for all users?
No. Claude identity verification is prompted for certain advanced capabilities, platform integrity checks, safety measures, and age restrictions. The rollout is gradual and affects select users globally. You may not encounter the prompt if you use Claude for basic tasks or if you are not in a targeted cohort.
Can you use Claude without submitting ID and a selfie?
Yes, but with limits. If you are prompted for verification and decline, you lose access to the gated features—but you may retain access to standard Claude capabilities. Anthropic has not specified which features remain available after declining verification. This ambiguity is a design flaw that should be clarified.
Does Anthropic use your biometric data to train Claude?
Anthropic states that Claude identity verification data is not used to train its models. However, the data is still collected, encrypted, and held by Persona. Not using it for training does not mean it is never analyzed for other purposes, such as abuse detection or user behavior research. The distinction matters.
Claude identity verification represents a calculated bet by Anthropic: that the compliance and safety benefits of gating features behind biometric verification outweigh the user friction and privacy concerns. For some users, that tradeoff is acceptable. For others—particularly those in regions where government ID verification carries geopolitical risk—it is a barrier that should not exist. The real test will come in the months ahead: does Anthropic’s system reduce abuse and regulatory headaches, or does it simply frustrate users and push them toward competitors who do not require a selfie?
This article was written with AI assistance and editorially reviewed.
Source: TechRadar
