Detailed Analysis
Anthropic quietly rolled out an identity verification system for Claude users around April 16, 2026, though the policy is significantly narrower in scope than the article's headline suggests. Rather than requiring all new users to submit a government-issued photo ID, the verification prompts are triggered only in a limited set of circumstances — primarily when Anthropic's systems detect potentially fraudulent or policy-violating activity, or when routine platform integrity and compliance checks flag an account. The rollout drew public attention after users began reporting unexpected verification prompts on social platforms, prompting Anthropic to address the policy through its Help Center documentation.
The mechanics of the verification process are handled by third-party identity provider Persona Identities. Users who encounter a prompt must submit a physical government-issued document — such as a passport, driver's license, or national ID card — along with a live selfie captured via phone or webcam. Digital IDs, photocopies, and non-government documents are explicitly rejected. The process is designed to be completed in under five minutes, with retry options available for technical failures such as blurred imagery. Anthropic retains the role of data controller in this arrangement, stipulating that collected data may only be used for verification, fraud prevention, and support purposes, and explicitly excluding it from AI training datasets. An appeals process is also available for users who believe their accounts were wrongfully flagged.
The policy reflects a broader tension in consumer AI platforms between accessibility and abuse prevention. As large language models become increasingly capable — and as Claude's user base expands — Anthropic faces mounting pressure to prevent misuse without creating friction that alienates legitimate users. The targeted, trigger-based nature of this verification system represents a deliberate middle path: rather than deploying blanket Know Your Customer (KYC) protocols at signup, the company is reserving intrusive identity checks for high-risk scenarios. This approach minimizes data collection across the general user population while still providing a deterrent and remediation mechanism for bad actors.
The move situates Anthropic within a growing industry-wide reckoning over platform integrity. Competitors including OpenAI and Google have faced criticism for the relative ease with which their AI systems can be accessed and misused at scale, and regulators in the European Union and elsewhere have begun scrutinizing AI providers' obligations around user verification, particularly in the context of the EU AI Act's risk-tiered compliance framework. Anthropic's choice to partner with Persona Identities — a specialized identity infrastructure company — rather than build verification in-house signals an awareness that compliance and trust infrastructure will become increasingly central to operating a consumer AI platform, even if the current implementation remains deliberately limited in scope.
Read original article →