AI company Anthropic is introducing identity verification checks for some users of its Claude platform. The change aims to improve safety, prevent misuse, and help meet legal requirements. Under the system, certain users may need to confirm identity before accessing some features. The policy takes effect July 8.
Anthropic is not running the checks itself. The company uses Persona, a third-party identity firm, to handle document and selfie verification. Users must provide a valid government-issued photo ID and a live selfie captured by phone or webcam. The process takes under five minutes and is designed to feel simple.
The rollout is limited and targeted, not universal. Anthropic says verification applies to a small subset of users whose accounts get flagged for potential policy violations. Those users gain a path to appeal through verification rather than losing access outright. Prompts may also appear during specific safety or compliance checks.
Accepted documents include passports, driver’s licenses, and national identity cards. Photocopies, screenshots, mobile IDs, and non-government IDs are rejected. Users should ensure their documents stay clear, undamaged, and unexpired before submitting.
On privacy, the picture is more nuanced than a simple deletion claim. Persona holds the ID and selfie, not Anthropic’s own servers. However, Anthropic can access verification records through Persona when needed, such as during an appeal. The company says it does not use verification data to train its models. Data is encrypted in transfer and at rest.
The move has drawn scrutiny too. Persona is backed by Founders Fund, which also invests in Anthropic. Critics note the live selfie generates facial geometry data, which carries higher privacy stakes under some state laws. If verification fails, users can retry with better lighting or a different ID. Anthropic says wrongly restricted accounts can appeal through its support process.
