Some Claude users may be asked to verify their identity with a government-issued document, and in some cases a live selfie. The reported checks are selective, not a new requirement for every Claude account, but they put identity documents in the path to using certain AI capabilities.
Anthropic Watch analysis
What happened
Coinpaper reports that checks may be triggered when users access certain capabilities, face a platform security review, or encounter safety and compliance requirements. It says Anthropic uses Persona to process documents, and accepts physical passports, driving licences and national identity cards. A live selfie may also be requested to match the user to their document.
The article says Anthropic describes the checks as a way to prevent fraud, enforce policies and meet legal obligations. It also says Anthropic remains the data controller, can access verification records through Persona when needed for matters such as account appeals, and says the data is not used to train Claude or for advertising.
Why it matters
This is a meaningful change in the relationship between an AI service and its users: access to some capabilities can involve sharing highly identifying information. A selfie and a passport are not just another box to tick, even when the request is limited to some accounts.
The distinction between selective checks and universal verification matters. So does the distinction between Anthropic holding data directly and a third party processing it on its behalf. Coinpaper says Persona stores the documents, while Anthropic can access records when necessary. Those are claims about the process, not a guarantee that sensitive information carries no risk.
Our read
Identity checks may help a service manage fraud and access, but asking for official documents raises a fair question about necessity and proportionality. If a check appears, users should pay attention to what triggers it, what data is collected and how an appeal works. “Only some users” is useful context; it is not the same as “nothing to consider”.
What to watch
- Which Claude capabilities or reviews trigger a verification request.
- How long Persona retains documents and what access Anthropic has in practice.
- Whether Anthropic gives users a clear alternative or appeal route when they cannot complete a check.
Discussion spark: Should an AI provider be able to require government ID or a live selfie for access to selected capabilities, or should those checks be limited to services where identity is essential?
Sources and evidence
- Source update (29 October 2026, 15:49 UTC)
Anthropic Watch is independently operated by WittyWires. It is not affiliated with, endorsed by, or operated by Anthropic.