Every high-demand launch, from concert tickets to token airdrops or limited edition sales, faces the same failure: bots and mass-buyers exploit scarcity faster than platforms can stop them. The AI Passport lets anyone running a launch (from a ticket drop to a limited edition product release or a token airdrop) verify that each buyer is one real, unique human, without collecting any data or seeing further information tied to the buyer’s identity.
Inspiration
Another concert on sale, another "sold out in 8 seconds" story, another wave of fans finding the same tickets on resale sites minutes later at triple the price. We started looking into why this keeps happening. Ticket drops, sneaker releases, token airdrops, even Indian Railways' Tatkal quota. Every one of these systems is trying to answer the same simple question - is this a real, unique human? - but none of them actually verify it directly. CAPTCHAs, purchase limits, queues, and allowlists are all just proxies for that question, and every proxy gets industrialized by bots faster than platforms can patch it.
The "obvious" fix was asking for an ID or a face scan - like the one by WorldID if you've seen the adverts around. However, it trades one problem for a worse one: now the platform is sitting on a pile of sensitive personal data it never wanted and now has to protect. We wanted to see if there was a way to answer the question without creating that liability at all.
What it does
Fair Launch is an AI Passport that lets anyone running a high-demand launch, like a ticket drop, a limited-edition release, a token airdrop, verify that each buyer is a real, unique human and hasn't already claimed their share, without ever seeing who that person is.
Think of it like a bouncer at a club door who can tell instantly whether someone is a real person and whether they've already used their one entry, without ever asking for a name, checking an ID, or writing anything down. The seller gets a yes/no answer. Nothing more.
How we built it
The system works in three steps:
- Local ledger check: Before generating anything, the passport checks its own encrypted local ledger against the drop's rules (e.g., previously bought 0, requesting 2, limit 5).
- Zero-knowledge proof: The passport proves the compound statement "unique human AND within allocation limit" as a single boolean, without revealing identity, purchase history, or the ledger itself.
- Stateless verification: The storefront checks the proof's math and grants checkout. It never stores a nullifier itself, so a second request past the limit is mathematically refused before it ever becomes a business decision.
Enrollment deliberately avoids biometrics. Instead of an iris or face scan, verification is derived from the depth of a user's existing cross-platform digital footprint: something a bot farm can't cheaply fabricate at scale. Enforcement is baked into the circuit itself: a fixed-limit drop mathematically restricts the passport to as many valid proofs as the limit allows, so a sixth attempt doesn't get rejected by a rule, it simply can't produce a valid proof.
Challenges we ran into
- Proving "uniqueness" without identity. The hardest design constraint was building something that could confidently answer "is this the same person who already claimed?" without ever knowing who the person is - every naive approach we sketched first leaked either identity or a persistent tracking handle.
- Avoiding the biometric trap. Face and iris scanning (World ID) solves uniqueness well but introduces hardware bottlenecks, adoption friction, and real regulatory risk - several jurisdictions have already pushed back on iris-scan models. We had to find a verification signal that didn't require special hardware or a physical checkpoint.
- Keeping it stateless for the seller. We wanted platforms to be able to verify a proof without having to run and secure their own nullifier database, which meant pushing more of the "memory" of past claims down into the passport's local, encrypted ledger.
Accomplishments that we're proud of
- Designed a verification flow that answers a genuine, provable yes/no question - one human, one entry, not yet claimed - without the seller ever collecting or storing personal data.
- Landed on an approach that avoids the three biggest weaknesses of existing proof-of-personhood systems: no biometric liability, no hardware bottleneck (no Orb, no physical checkpoint), and no single point of centralization.
- Validated the problem directly with an industry practitioner - Rodrigo Bautista, CEO of FenEx - who confirmed that fair access and duplicate-purchase prevention is a real, unsolved pain point, while flagging adoption and ecosystem integration as the real hurdle (which shaped our go-to-market plan).
- Scoped a realistic, sequenced path to a real pilot rather than trying to boil the ocean on day one.
What we learned
The gap in every existing anti-bot defense isn't a lack of effort - it's that none of them verify the actual fact that matters ("this is one unique person") directly. They all approximate it, and approximations get gamed. We also learned that "more verification" isn't automatically better: the moment you ask for an ID or a face scan, you've traded a bot problem for a data-liability problem, which is exactly the failure mode that has landed centralized proof-of-personhood systems in regulatory trouble. The right solution has to prove the fact and forget everything else in the same step.
What's next for Fair Launch
- Prove the mechanism: Finalize the enrollment and nullifier design, then build a local-ledger + zero-knowledge-proof demo around one simulated ticket launch.
- Pilot with one real partner: Run it live on a small-scale launch with a willing independent artist or project, and measure bot capture before and after for real evidence.
- Earn the right to scale: Take pilot results to larger ticketing, blockchain, and retail platforms already spending on weaker bot defenses. Because verification runs client-side, it costs us very little to operate - platforms would pay per verification or per launch, less than what they spend today on defenses that don't work.
Beyond ticketing, the same one-human-one-entry primitive applies directly to sneaker and luxury drops, NFT mints and token airdrops, limited in-game item releases, retail IPO allocations, and public-sector lotteries - anywhere scarcity currently gets captured by whoever can run the most bots.
Built With
- circom
- node.js
- nullifiers
- oauth
- react
- semaphore
- snarkjs
- typescript
- webassembly
- zero-knowledge-proofs
- zk-snarks

Log in or sign up for Devpost to join the conversation.