Answers are grounded in this paper and cite their sources.
More than half of the traffic on the internet now comes from machines, and the web has no reliable way left to tell a person from a program. CAPTCHAs have been beaten, an agent driving a real browser from a home connection looks like anyone else, and every rule we might write about agents rests on a distinction the internet can no longer make. This paper takes up how a person proves they're a person without the internet learning who they are. The answer has two layers. The first is a strong binding done once: a personhood credential issued by an institution that has looked a person in the face, built so that it can be used over and over without the uses being linked, with issuers now arriving, led by the European Union's digital identity wallet. The second is a cheap check done often: a few seconds of proof that a real person is present right now, taken from the phone in nearly every pocket, signed by its hardware, and returned as a token that says a verified person vouches for this action and nothing more. Three rules hold the design to the right standard. The gate learns that a person is there and never who. The checks run in short bursts when a decision is pending and never as a continuous feed, because a stream of motion or pulse data is a health record. And the system has to work for the people a naive design turns away, including people with Parkinson's or a prosthetic, people who share a phone with a caregiver, and people with no phone at all. Proving personhood can make impersonation expensive and hard, and it can never make it impossible. Humanity Systems working paper. Part of a series on one question: how do we trust AI, now and into the future?
No takes yet. Share an insight, caveat, or question.
A TUCKER (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: