Anthropic Now Wants Your Passport
Anthropic just started asking some Claude users for a government ID and a selfie. Not a password, an actual passport or driver's license, run through Persona, about five minutes. Fail it, or trip a policy check, and your account can get banned. This is on the official support page that hit the top of Hacker News today with 449 points.
Right now it's scoped to a few use cases, what Anthropic calls routine platform-integrity checks, but the direction is unmistakable. The lab that builds Mythos-class models, the same models the government wanted restricted earlier this month, is moving toward knowing exactly who is on the other side of the API. For high-capability work, anonymous access is ending.
Why does this matter beyond the privacy headache? Because it's the clearest sign yet that frontier AI is being treated like a regulated utility. Banks do KYC because money moves through them. Anthropic is now doing KYC because capability moves through it, the ability to write exploit code, design bio sequences, run autonomous agents at scale. When the tool is powerful enough to cause real harm, who are you stops being optional.
The agent angle is the sharp one. As people wire Claude into autonomous agents that act on the world, the identity question gets thornier. Whose ID backs an agent that runs for hours, spends money, touches systems? Anthropic verifying the human behind the account is step one of a longer story about who answers for what agents do. Persona keeps the data, Anthropic says it won't train on it, but the precedent is set. The most capable AI now comes with a turnstile.
Link: https://support.claude.com/en/articles/14328960-identity-verification-on-claude
← Back to all articles
Right now it's scoped to a few use cases, what Anthropic calls routine platform-integrity checks, but the direction is unmistakable. The lab that builds Mythos-class models, the same models the government wanted restricted earlier this month, is moving toward knowing exactly who is on the other side of the API. For high-capability work, anonymous access is ending.
Why does this matter beyond the privacy headache? Because it's the clearest sign yet that frontier AI is being treated like a regulated utility. Banks do KYC because money moves through them. Anthropic is now doing KYC because capability moves through it, the ability to write exploit code, design bio sequences, run autonomous agents at scale. When the tool is powerful enough to cause real harm, who are you stops being optional.
The agent angle is the sharp one. As people wire Claude into autonomous agents that act on the world, the identity question gets thornier. Whose ID backs an agent that runs for hours, spends money, touches systems? Anthropic verifying the human behind the account is step one of a longer story about who answers for what agents do. Persona keeps the data, Anthropic says it won't train on it, but the precedent is set. The most capable AI now comes with a turnstile.
Link: https://support.claude.com/en/articles/14328960-identity-verification-on-claude
Comments