Summary

Anthropic is implementing identity verification for Claude users and requiring the submission of government-issued identification documents and, in some cases, a live selfie before accessing certain platform features. The verification process is handled by third-party provider Persona Identities and typically takes less than five minutes. Anthropic emphasizes that verification data is not used for training AI models and is used exclusively for identity confirmation and compliance with legal and security obligations. The company positions itself as one of the first major consumer AI providers with formal identity verification as an access control mechanism.

People

  • Pareesa Afreen (Reporter, The News)

Topics

  • Artificial Intelligence & Identity Verification
  • Data Protection and User Data
  • Platform Integrity and Abuse Prevention

Clarus Lead

The introduction of identity verification by Anthropic signals a turning point in the regulation of consumer AI platforms: while the industry has largely operated without verification barriers, Anthropic is now implementing access controls to prevent abuse and meet legal obligations. This reflects growing regulatory pressure and potential accountability requirements that large language model providers will need to meet in the future. At the same time, the lack of transparency regarding affected features heightens user questions about the proportionality of this measure.

Detailed Summary

Anthropic is partnering with Persona Identities, a specialist in digital identity infrastructure. The platform only accepts physical government-issued photo identification documents – passports, driver's licenses, national ID cards, or government-issued credentials – from most countries. Photocopies, screenshots, digital IDs, student IDs, and bank cards are explicitly rejected. The selfie element serves as a liveness check.

According to Anthropic, data management follows strict restrictions: the company has access to verification records residing on Persona's system but does not create additional copies. Central to this is the assurance that verification data does not enter model training – a key differentiator from other AI providers. Data sharing with third parties occurs only for verification and compliance purposes. Anthropic frames the measure as integrity assurance rather than a product feature: "Being responsible with powerful technology starts with knowing who is using it" states the official announcement. The rationale includes abuse prevention, policy enforcement, and legal obligations. What remains unclear is which specific features trigger verification – Anthropic only refers to "certain capabilities" and "routine integrity checks."

Key Findings

  • Anthropic requires government-issued identification documents and live selfies for verification – a first among major consumer AI providers
  • Verification data is not used for model training and is subject to strict data protection
  • The affected features/functions remain publicly unspecified; Anthropic only mentions "certain capabilities"

Critical Questions

  1. Data Quality & Source Validity: Anthropic confirms the verification practice only via support post – is there independent documentation of the technical implementation by Persona?

  2. Conflicts of Interest: Persona Identities is a commercial provider – how transparent are the contracts between Anthropic and Persona regarding data access and deletion?

  3. Scope & Proportionality: Anthropic mentions "certain capabilities" – does verification cover all new features, only high-risk applications, or only a pilot program? Clear public documentation is lacking here.

  4. Data Security: How long does Persona store selfies and ID images? What encryption protects this highly sensitive data in transit and at rest?

  5. Alternatives: Why does Anthropic require physical selfies instead of document-based verification alone – is this technically necessary or a risk assessment decision?

  6. Industry Standard: To what extent will this verification become standard for other Claude competitors (OpenAI, Google), or will it remain Anthropic-specific?

Source Directory

Primary Source: Claude now wants your passport: Here's what Anthropic does with it – The News, April 17, 2026

Verification Status: ✓ April 17, 2026

This text was created with the assistance of an AI model. Editorial Responsibility: clarus.news | Fact-Check: April 17, 2026