← back to blog

Proof of personhood checks and who they lock out

Why projects reach for proof of personhood

An airdrop budget is a fixed pool split among however many wallets show up and qualify. If a script can generate a thousand wallets and pass every eligibility filter a project sets, the reward per genuine user drops accordingly. Most sybil defenses try to raise the cost of that thousand-wallet script through activity heuristics: minimum transaction counts, holding periods, gas spent, cross-wallet funding patterns. Proof of personhood tries something more direct. Instead of inferring humanness from behavior, it asks a wallet to prove it’s tied to exactly one human being, verified once, before anything else counts.

That’s a fundamentally different claim than “this wallet looks organic.” A wallet can look organic and still be one of fifty controlled by the same operator. Proof of personhood is meant to close that gap by putting a floor under the identity layer itself, not the activity layer.

How the major approaches actually work

Biometric scanning. World ID (formerly Worldcoin) uses a device called the Orb that scans a person’s iris and generates a unique code, which it claims is derived locally and not stored as a raw image. The person gets a credential they can attach to a wallet and reuse across apps that support World ID. Humanity Protocol has pursued a similar model using palm scans instead of iris.

Social graph verification. BrightID doesn’t use biometrics at all. It asks users to do a video verification party with other existing, previously-verified BrightID users, building a web of vouching connections. The system then runs graph analysis to spot clusters that look artificially connected, the social equivalent of the wallet-clustering heuristics chain analysts run on-chain.

Staked vouching. Proof of Humanity works on a registry model: a person submits a short video and a profile, and existing registered humans can vouch for them or challenge the submission with a bond at stake. Fraudulent entries can be challenged and removed, with the challenger collecting part of the bond.

Puzzle-based liveness. Idena takes a different angle with a synchronized “flip test,” a CAPTCHA-style challenge every participant solves at the same scheduled time worldwide, on the theory that scripting a bot to solve thousands of these in the same narrow window is harder than scripting ordinary form submissions.

Aggregated stamps. Gitcoin Passport doesn’t verify personhood on its own. It aggregates signals from other accounts, wallet age, Coinbase/BrightID/GitHub verification, ENS ownership, and produces a composite score projects can set a threshold against. It’s a reputation score built from other people’s verification work, not a standalone check.

Government ID plus selfie. Providers like Fractal or Civic compare a photo ID against a live selfie, similar to KYC flows at exchanges. This is the strictest in terms of legal identity binding, and the one with the fewest fallback options if it fails.

Each of these picks a different tradeoff between privacy, friction, and how hard the check is to fake at scale. None of them is free of exclusion.

Who actually gets locked out

People without access to the hardware. Orb locations for World ID are concentrated in specific cities, which means verification often requires travel that isn’t practical or affordable for people in rural areas or countries where no Orb operates. Palm and face-scan systems assume a phone with a decent camera, which not everyone owns.

People who won’t hand over biometric data. Even with local-processing claims, submitting an iris or palm scan to a company’s infrastructure is a permanent decision that can’t be undone if the person changes their mind later. A meaningful number of otherwise eligible users decline on principle, and for a project that gates claims behind that one method, declining means being excluded entirely.

People who fail liveness checks for reasons unrelated to being human. Liveness detection (blink, turn your head, follow a dot) is tuned on specific camera quality and lighting assumptions. It disproportionately fails people on older phones, people in low-light housing, and people on slow connections where video frames drop. It can also fail people with certain motor or vision impairments who can’t reliably perform the requested motion on cue.

People without ID that matches the system’s expectations. Undocumented people, refugees, people who’ve legally changed their name, and people in countries with weak civil registries all run into the same wall with ID-plus-selfie systems: the check assumes a clean, current, machine-readable government document exists and matches the face in front of the camera. When it doesn’t, there’s usually no manual appeal path that resolves quickly.

People flagged by network signals during verification. Some of these flows layer in IP or device fingerprinting on top of the biometric or ID check, flagging sessions coming from VPNs, corporate NAT, or shared connections. That catches privacy-conscious users, people on restrictive networks, and travelers just as often as it catches anyone actually trying to fake the check.

People building legitimate reputation from scratch. Systems like BrightID and Proof of Humanity depend on already-verified people vouching for new ones. That works fine once a network is established, but it creates a cold-start problem for anyone who doesn’t already know someone in the system, which skews participation toward people already embedded in crypto-native social circles.

Where this connects to wallet-level defenses

It’s worth being clear-eyed about what proof of personhood is actually built to stop: one person claiming an airdrop allocation many times over by controlling many wallets. That’s a distinct target from the funding-graph and timing heuristics chain analysts use to cluster wallets after the fact. A wallet that passes an on-chain sybil filter by having varied transaction history, separate funding sources, and no obvious clustering signature can still fail proof of personhood if it’s tied to a human who’s already claimed under a different identity, because the check operates at the identity layer, not the activity layer.

That matters for anyone running multiple wallets for legitimate reasons, portfolio separation, testing, different risk profiles, since none of that changes the fact that proof of personhood systems are explicitly designed to cap participation per human being, not per wallet. Routing around that cap by using someone else’s verified credential, buying a verified account, or splitting one person’s identity across accounts isn’t a wallet-privacy technique, it’s the exact behavior these systems exist to catch, and most of them treat it as a terms-of-service violation with account-level consequences, separate from any chain-analysis risk the wallets themselves carry.

The practical takeaway: passing a proof of personhood check and passing on-chain sybil heuristics are two separate hurdles, and clearing one says nothing about the other. A wallet still gets evaluated on funding patterns, gas behavior, and timing by any exchange, protocol, or analytics firm that looks at it later, regardless of what identity credential unlocked the original claim. Treating proof of personhood as a box to check rather than a real constraint on how many claims one person can make is how accounts end up flagged well after the airdrop is over.

If you’re trying to understand how airdrop eligibility actually gets decided, and how the wallet and identity layers each get scrutinized, more breakdowns like this live on the site homepage.

Get new guides and videos first — join the Telegram channel.

need infra for this today?