What is proof of personhood?
What happens if, in an online popularity poll, one person creates 1,000 fake accounts and votes with all of them? The result is easily distorted. As AI becomes able to behave just like a human, this problem becomes more serious. Proof of personhood is a way of showing that "behind this account is one real human being".
In short
Proof of personhood is a way of showing online that "this account belongs to a real human, and that human has only one".
It reveals no name or address. It proves only two things: that you are human, and that you are a different person from everyone else. The aim is to keep votes and handouts of money from being taken over by someone who can create fake accounts in bulk.
For example
World (formerly Worldcoin). OpenAI co-founder Sam Altman and others launched it in 2019. A dedicated device called the "Orb" scans the iris (the pattern in the eye) and turns it into a code that cannot be turned back (a hash). Later scans are checked against it, to stop the same person registering twice [1]. The 2023 version of its white paper (the explanatory document) says that as AI grows more powerful, it becomes harder to tell humans from bots, so this mechanism will become an essential basic building block of the digital world, needed for online voting and large-scale handouts [2]. In 2025, Vitalik Buterin, the founder of Ethereum, noted that World's ID had more than 10 million users [3].
The Orb comes to Tokyo. The Orb's world tour began in Tokyo in April 2023. Around the same time, a developer event in Tokyo held a session on global proof of personhood in the age of AI. Hirokazu Nishio's notes quote the announcements from that time [4].
Methods based on vouching by acquaintances. There are other methods too: uploading a video of your face and having existing users vouch for you; checking each other on video calls; a quiz everyone solves at the same set time; and existing users declaring that they trust a person. Some of these methods were combined and used to decide which accounts counted as valid in Quadratic Voting [5].
How it works
What it proves
In a 2023 article, Buterin gives the simplest definition [6]: creating a list that guarantees each key (something like an account's password) is controlled by a different human being.
- A human can put one key on the list.
- A human cannot put two.
- A bot cannot put any.
Buterin explains that this is a limited form of a real-world identity document, one that ideally does not reveal which person it is [7].
Why it is needed
Without it, Buterin says, ways of deciding things together online (including small ones like voting on social media posts) are easily taken over by the very rich or by hostile governments [6].
A 2020 paper by the researcher Divya Siddarth and others also points out that many ways of making decisions in the cryptocurrency world decide who may take part by how much money or computing equipment they hold, and so become rule by the wealthy (plutocracy). For social and political uses, the paper argues, a way of confirming one human being is needed [8].
Two families
Broadly, there are two approaches [9]:
- Checking by body information (biometrics). World's iris scans, for example.
- Checking by social connections. For example, four people who have already been verified say this person is human.
Buterin accepts that the social-connection approach has advantages: it needs no special hardware, it collects no body information, and it can express whether someone is human as a score rather than a yes-or-no. But he also lists weaknesses: you cannot get in unless you already know someone (whole regions can be left out), relationships can leak, and people with many connections can create more IDs. Buterin's conclusion is that there is no ideal method. He recommends a combination: start with biometrics and supplement it with social connections [9][10].
How strict should "one per person" be?
World's 2026 white paper argues that the strictness of exactly one per person is essential [11]. An attack is possible in which a small number of people pass registration many times and lend their credentials to bots. The white paper offers a calculation: if 1% of the US population each sold 9 of 10 credentials, a single attacker could pass as 31 million Americans. It concludes that because social connections can be imitated by AI, only biometric methods can meet the requirements at global scale [12].
The opposing view
In 2025, Buterin put forward a different view [3][13]. Wrapping an ID in cryptography that shows only that a condition is met, without revealing the contents, still leaves most of the dangers, he says: loss of privacy, being forced to show your ID, and being judged wrongly. Those dangers, in his view, come from trying to enforce one per person in the first place.
On the other hand, using money to deter fake accounts (paying the same price for each ID) favors the rich too much. Buterin writes that the theoretical ideal is N identities at a cost of N². (Worked example: under this formula, if one ID costs 1, three cost 9 and ten cost 100. The numbers are this article's own illustration.) In other words, the more you hold, the sharply more expensive each one becomes. Buterin thinks the closest real-world answer is pluralistic identity: an identity system that no single issuer dominates [13].
What is debated
- World versus the book and Buterin. World argues that strict one-per-person and biometric methods are the only answer [11][12]. The book ⿻ Plurality and Buterin see relying on a single method or a single issuer as dangerous in itself [1][3]. Still, in 2023 Buterin also acknowledged that World is considerably better at protecting privacy than methods that publish faces or publish social relationships. He asks readers to separate problems specific to World, problems common to biometric methods, and problems common to this whole kind of mechanism [14].
- Stand-ins can be hired. A 2022 paper by Buterin and others points out that while most people have not registered, an attacker can hire people who have not yet registered to act as stand-ins. So even with this mechanism, the paper says, fake-account attacks become only somewhat more expensive [15].
- Ethical criticism. In 2022, the US magazine MIT Technology Review reported deception and exploitation in how users were recruited during Worldcoin's early trials [16]. Buterin also refers to this reporting [7].
Common misunderstandings
- A common misunderstanding. Proof of personhood does not equal biometric authentication. Biometrics is only one family; there are also methods based on social connections, and combinations of the two [5][10].
Where it fits in Plurality
In chapter 4-1, the book Plurality writes that what an identity system can offer "depends intimately on how richly it can establish user identity" [17]. "Systems that can determine a user is unique but nothing else can only offer services that can legally and practically be made available to every person on the planet" [17].
The book then criticizes Worldcoin. It points out that it is "vulnerable to stealing or selling of accounts"; that it cannot show what most services need, such as a name, citizenship or educational credentials; and that if irises can one day "be spoofed by artificial intelligence systems combined with advanced printing technology", it "may be subject to an extreme 'single point of failure'" [1]. It sums up: "biometric systems are too reductive to establish and protect identities with the richness and security required to support ⿻" [1].
The book's answer is Identity and Personhood (⿻ identity), which builds identity out of the overlap of many social relationships. Proof of personhood corresponds to one property within it: being a unique human. The 2022 paper by Buterin and others also says it aims for an overlapping foundation that moves from being a single unique human toward being each distinct human being [15].
At the same time, deterring fake accounts is a precondition for Quadratic Voting, Quadratic Funding and ⿻ Voting. The book writes: "Without a reasonably secure identity foundation, any voting system, for example, will be captured by whoever can produce the most false credentials, degenerating into a plutocracy" [17].
Further reading
- Plurality, chapter 4-1 "Identity and Personhood" (English, book) — https://www.plurality.net/v/chapters/4-1/eng/
- Vitalik Buterin, "What do I think about biometric proof of personhood?" (English, article) — https://vitalik.eth.limo/general/2023/07/24/biometric.html
- World white paper 2, "Achieving Proof of Human" (English, white paper) — https://whitepaper.world.org/achieving-proof-of-human
- Siddarth and others, "Who Watches the Watchmen?" (English, paper) — https://arxiv.org/abs/2008.05300
- Ohlhaver, Weyl and Buterin, "Decentralized Society: Finding Web3's Soul" (English, paper) — https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4105763
- Vitalik Buterin, "Does digital ID have risks even if it's ZK-wrapped?" (English, article) — https://vitalik.eth.limo/general/2025/06/28/zkid.html
Sources
- Plurality, 4-1, section "Public and decentralized digital identity" — https://www.plurality.net/v/chapters/4-1/eng/ [CC0]
- Worldcoin white paper (2023 version) — https://whitepaper.worldcoin.org/
- Vitalik Buterin, "Does digital ID have risks even if it's ZK-wrapped?" (2025) — https://vitalik.eth.limo/general/2025/06/28/zkid.html
- Hirokazu Nishio, "Worldcoin Orb and Tokyo" (Cosense) — https://scrapbox.io/nishio/Worldcoin_Orb_and_Tokyo
- Vitalik Buterin, "What do I think about biometric proof of personhood?", on early attempts (2023) — https://vitalik.eth.limo/general/2023/07/24/biometric.html
- Same, section "What is proof of personhood and why is it important?" — https://vitalik.eth.limo/general/2023/07/24/biometric.html
- Same, opening — https://vitalik.eth.limo/general/2023/07/24/biometric.html
- Siddarth, Ivliev, Siri and Berman, "Who Watches the Watchmen?" (2020) — https://arxiv.org/abs/2008.05300
- Vitalik Buterin, "What do I think about biometric proof of personhood?", comparison of biometric and social-graph methods — https://vitalik.eth.limo/general/2023/07/24/biometric.html
- Same, section "So... what do I think?" — https://vitalik.eth.limo/general/2023/07/24/biometric.html
- World white paper 2, "Achieving Proof of Human", "Importance of Uniqueness" (2026) — https://whitepaper.world.org/achieving-proof-of-human
- Same, "Web of Trust" and "Biometrics" — https://whitepaper.world.org/achieving-proof-of-human
- Vitalik Buterin, "Does digital ID have risks even if it's ZK-wrapped?", on N² cost for N IDs — https://vitalik.eth.limo/general/2025/06/28/zkid.html
- Vitalik Buterin, "What do I think about biometric proof of personhood?", on problems with Worldcoin's design — https://vitalik.eth.limo/general/2023/07/24/biometric.html
- Ohlhaver, Weyl and Buterin, "Decentralized Society: Finding Web3's Soul", section 8.3 (2022) — https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4105763
- MIT Technology Review, reporting on Worldcoin (2022-04-06) — https://www.technologyreview.com/2022/04/06/1048981/worldcoin-cryptocurrency-biometrics-web3/
- Plurality, 4-1 "Identity and Personhood", opening — https://www.plurality.net/v/chapters/4-1/eng/ [CC0]
We welcome error reports and questions ("What is…?"). Submissions become public issues.