Anthropic Implements ID Verification to Restrict Access

Anthropic has started rolling out identity verification for a small set of Claude users, requiring a government-issued photo ID and, in some cases, a live selfie before permitting access to certain capabilities. The checks are executed by third-party provider Persona, and Anthropic says verification is applied selectively for platform integrity, fraud or abusive-behavior signals, and to satisfy legal obligations. Reports also indicate the measure is intended to limit access from some jurisdictions tied to U.S. national security concerns, including China, Russia, and North Korea.

Anthropic routes verification through Persona rather than ingesting raw documents directly into its systems. Accepted credentials include passports, driver's licenses, and national ID cards; screenshots, photocopies, and non-government IDs are rejected. Anthropic says Persona stores and processes the images under contractual limits, encryption in transit and at rest, and specified retention policies, and that verification data will not be used to train models. The verification prompt is applied conditionally, appearing when users attempt specified features or when automated signals flag potentially fraudulent or abusive activity.

• •Identity gating changes threat models for platform abuse, shifting some mitigation from content-based controls to identity-linked accountability.
• •Integrations with identity providers like Persona create new data-handling and compliance dependencies; review of retention, audit logs, and access controls is essential.
• •Models and telemetry teams should expect reduced anonymity for a subset of users, which will affect A/B tests, behavioral analytics, and opt-in research cohorts.

This is one of the first major consumer-facing generative AI platforms to adopt formal identity verification as a gating mechanism for features. The move intersects three trends: stronger platform integrity controls post-misuse incidents, geopolitical export-control and supply-chain concerns, and growing industry experimentation with identity-linked access for higher-risk capabilities. Anthropic frames the change as safety-first; critics note privacy tradeoffs, especially given Persona investors like Founders Fund and connections between surveillance-capital firms and government contractors such as Palantir. That provenance has amplified user distrust despite Anthropic's assurances that ID images will not feed model training.

Whether competitors adopt similar ID-gating for high-risk features, how regulators treat identity-based controls for AI, and whether contractual protections and audits around third-party verifiers are sufficient to maintain user trust. Also monitor appeals, false-positive rates on gating signals, and any policy or technical updates that narrow or expand which Claude capabilities require verification.