Why does a currency need proof of human?
Any system that gives every human a daily allocation has one obvious attack: pretend to be many humans, collect many allocations. This is the sybil problem, and it decides whether the entire design works.
AI makes it urgent for everyone else too. When bots can write, talk, and transact indistinguishably from people, "is this a real person?" stops being a philosophical question and becomes an economic one. Money that cannot tell humans from bots will be farmed by bots.
What are the two problems, exactly?
Proof of human splits into two distinct questions that need two distinct kinds of answers.
Is this a real human? This is a judgment call on evidence quality, so it is handled by human judgment: miners review documents, biometrics, behavioral signals, and vouches, and assign a score.
Does this person have only one account? This is a math and pattern problem, so it is handled at the protocol level: cryptographic nullifiers ensure the same credential can never open two accounts, and transaction-pattern analysis flags lives that don't add up (an account transacting in person in Tokyo and Buenos Aires in the same afternoon is not one human).
How does verification actually work?
Every new account is reviewed by a panel of three independent miners, assigned by a first-in, first-out queue so nobody can pick their own reviewer. Each miner scores the evidence independently; the account's initial score is the median of the three.
Evidence is flexible on purpose. Government ID, biometric scans, in-person transaction confirmations, and social vouching all count, but one guarantee is hardcoded into the protocol: a person can reach 100% through social vouching alone. Other verified humans stake a percentage of their own savings as collateral on your legitimacy, and if you turn out to be fraudulent, their stake burns. Because the stake is a percentage of holdings rather than a fixed amount, a poor person's friends can vouch exactly as effectively as a rich person's.
Why hardcode the vouching path? Because if government ID were required, any government could unbank anyone by restricting issuance. The system verifies humans, not citizens.
Privacy is structural, not promised. No ID, biometric, photo, or personal record is ever stored on the blockchain. Miners record only a cryptographic hash proving a review occurred. The chain knows that evidence was reviewed; it never knows what the evidence was.
Why score the account and not the points?
Because the alternative has already been tried and it breaks money. If trust attaches to the points themselves, then a point's value depends on who issued it and who is looking at it. That is observer-relative money: your points are worth 100 to your friends and 60 to a stranger. Circles built exactly this, and the result was a currency merchants couldn't price goods in.
Scoring the account keeps every point identical. A point from a 40% verified account and a point from a 100% verified account are the same point once spent; the discount was applied at the account level, at the moment of spending. Merchants price in one unit. Fungibility survives.
What stops fraud after verification?
Enforcement, funded by economics rather than altruism. Any miner can challenge a suspicious account by staking points. An 11-miner jury, randomly selected, hears the case. If the account is condemned, the challenger collects 20% of its Earned balance as bounty and the remaining 80% burns. If the challenge fails, the challenger's stake is forfeited. Fraud-hunting pays; frivolous accusation costs.
Verification also decays about 10% per month, refreshed by ordinary life: a handful of in-person transactions per month fully offsets it. Being human is not a one-time exam. It is an ongoing pattern that is cheap to live and expensive to fake.
Does it need to be perfect?
No, and this is the honest framing. Existing systems already tolerate significant fraud: credit card networks, government benefits, ad networks full of bots. The bar is not zero fraud. The bar is making fraud more expensive than it is in the systems we already accept, and the combination of enrollment fees, staked vouches, decaying scores, pattern analysis, and bounty-hunting miners is built to clear it.
FAQ
Is my biometric data stored on the blockchain? No. Raw evidence never touches the protocol. Only cryptographic hashes of review events are recorded.
What happens if I'm wrongly accused? The burden of proof is on the challenger, you may appeal once to a fresh jury, and a resolved case grants a six-month protection window against re-challenge.
How is this different from Worldcoin? Worldcoin answers "is this a human?" with a single method (iris biometrics) and attaches a conventional token. The Alignment Economy accepts many evidence paths, requires none of them individually, and fuses verification into the money itself. Full comparison here.