← Back to Patent Claims
Patent Claim 108 All Patents →

Biometric Proof-of-Humanity with On-Device ZK Uniqueness Proof

Multi-Modal On-Device Biometric Capture, Zero-Knowledge Uniqueness Proof, and a Soulbound Non-Transferable Humanity Token

Patent Claim JIL Sovereign July 2026 Claim 108 of 157

Executive Summary

JIL Sovereign's Biometric Proof-of-Humanity (BPoH) system is a protocol-native identity primitive that lets a wallet holder prove they are a unique, living human being without any biometric data - raw or derived - ever leaving their device or becoming visible to JIL, a validator, or any counterparty. Rather than bolting a third-party KYC vendor onto an existing chain, BPoH embeds human verification as a first-class on-chain primitive: a single boolean any contract can check.

Three innovations distinguish this architecture: (1) multi-modal biometric capture and feature extraction that executes entirely on the user's device, so raw biometric data is architecturally incapable of reaching a server; (2) a locally-generated one-way commitment plus a zero-knowledge proof that the underlying biometric has never previously enrolled anywhere on the network, submitted in place of the biometric itself; and (3) a soulbound (non-transferable) ERC-721 Humanity token, minted only after both the uniqueness proof and an independent biometric-commitment check clear a standalone global registry.

Core Innovation: On-device-only biometric processing combined with a chain-verifiable, replay-proof uniqueness commitment and a transfer-disabled token - Sybil resistance becomes an L1-native primitive that proves uniqueness while learning nothing about the person behind it.

Problem Statement

Most blockchain ecosystems are, at the protocol level, wallet-based rather than human-based. A single actor can generate an unbounded number of wallets at near-zero marginal cost, and the network has no native way to distinguish one human controlling one wallet from one human controlling ten thousand. This gap is exploited systematically: bots and Sybil clusters have captured outsized shares of token airdrops and public-goods distributions, governance votes are diluted by holdings fragmented across throwaway wallets, and a large share of on-chain "social" activity is automated rather than human.

$10B+
Estimated annual cost of Sybil/bot exploitation across crypto ecosystems (industry estimate)
up to 99%
Share of value captured by bot/Sybil wallets in documented public airdrop cases
0
Bytes of raw biometric data transmitted off the enrolling device under BPoH

Why Existing Solutions Are Insufficient

  • Vendor-hosted biometric KYC (Persona, Jumio, Onfido-style): biometric templates are captured and matched on vendor infrastructure. A single vendor breach exposes templates for every enrolled user, and the verification result is scoped to that vendor's own database, not to the chain.
  • Hardware-orb proof-of-personhood networks: require dedicated capture hardware and centralize raw iris/biometric data at enrollment kiosks operated by the network or its partners.
  • Social-graph webs-of-trust: infer uniqueness from vouching relationships rather than biometric evidence, which is itself gameable at scale by coordinated ring vouching.
  • Simple soulbound "verified" badges without a uniqueness proof: can mark a wallet as attested but cannot cryptographically stop the same human from enrolling a second wallet, because there is no global commitment tied to the underlying biometric itself.

Technical Architecture

Multi-Modal On-Device Capture

Biometric modalities are combined so that no single sensor becomes a single point of failure. Feature extraction and hashing occur entirely on-device; nothing below is ever transmitted in raw form.

ModalityMethodFalse-Accept RateCapture Time
3D Face RecognitionDepth-map scan via structured-light / TrueDepth-class sensor~1 in 1,000,000~30s
Voice PrintVocal-tract resonance, formant structure, deepfake detection~1 in 100,000~10s
Behavioral BiometricsTyping rhythm, touchscreen pressure, gesture dynamics~1 in 50,000~2 min
Liveness DetectionRandomized challenge-response, skin micro-texture analysisSupplementary gate~5s

Enrollment Pipeline

  1. Device sensors collect the enabled modalities; on-device ML models extract feature vectors.
  2. An irreversible one-way hash (biometricHash) is generated locally from the extracted features.
  3. A zero-knowledge proof is constructed on-device attesting "this hash belongs to a unique human who has not previously enrolled" (uniquenessProof), plus a separate liveness attestation (livenessProof).
  4. Only the hash, the two proof digests, and a 0-100 confidence score are submitted on-chain - never the underlying biometric.

Smart Contract Layer

ContractRole
BiometricHumanitySoulbound ERC-721 enrollment and proof storage; stores the HumanityProof struct (biometricHash, uniquenessProof, livenessProof, enrollmentTime, humanityScore, lastVerification, verified, revoked) per wallet.
UniquenessRegistryStandalone contract, deliberately separated from BiometricHumanity so the deduplication set survives contract upgrades; tracks consumedProofs and enrolledBiometrics as two independent dedup keys.
HumanityGateImportable modifier library; a single requireHumanity(minScore) modifier gates any downstream contract function behind a minimum verified-human confidence score.

On enrollment, the contract verifies an ECDSA signature from an authorized VERIFIER_ROLE attestor over a digest binding the biometric hash, uniqueness proof, humanity score, block.chainid, and the contract's own address - preventing the signature from being replayed against a different contract or a different chain. Only after signature verification does the contract call the registry's consumeProof (reverts if the ZK proof digest was already used) and enrollBiometric (reverts if the biometric hash itself was already enrolled) - two independent, atomic dedup checks - before minting the token.

Attack Resistance & Reverification

AttackTechniqueDefense
Deepfake videoAI-generated face video3D depth requirement; unpredictable liveness challenges; skin micro-texture analysis
Photo replayHigh-resolution photo to cameraNo blink/head-movement/depth signal; behavioral biometrics absent
Synthetic voiceTTS or voice cloningVocal-tract mismatch; micro-timing absent; spectral synthesis detection
3D maskSilicone or 3D-printed maskNo pores/blood-flow signal; sub-dermal checks; behavioral mismatch
Duplicate enrollmentSame person, new walletHash collision rejected by UniquenessRegistry
NFT transferBuy or steal a Humanity tokenSoulbound: transfer function reverts unconditionally

A needsReverification() view compares block.timestamp against lastVerification + reverificationPeriod (default 365 days; governance may only raise the floor to at least 30 days). Revocation, exercised by a governance-held REVOKER_ROLE, checks the revoked flag before the verified flag specifically so a revoked wallet remains distinguishable from a never-enrolled one rather than collapsing to the same error state.

Prior Art Differentiation

ApproachBiometric leaves device?Dedicated hardware required?Cryptographic uniqueness proof?Non-transferable token?
Vendor-hosted KYC (Persona/Jumio-style)YesNoNoNo
Hardware-orb proof-of-personhoodYes, to enrollment kioskYesPartialVaries
Social-graph webs-of-trustN/A (no biometric)NoNoVaries
Generic soulbound "verified" badgeN/ANoNoYes
JIL BPoHNoNoYesYes

Patent Claim

Independent Claim 108: A computer-implemented system for privacy-preserving proof of human uniqueness, comprising: an on-device biometric capture and feature-extraction module configured to process at least one biometric modality entirely on a user's device and to output only a one-way cryptographic commitment of the extracted features, the underlying biometric data never being transmitted from the device; a zero-knowledge proof generator, executing on the device, configured to construct a proof that the commitment corresponds to a human who has not previously enrolled, without revealing the commitment's derivation; a uniqueness registry smart contract, separate from an enrollment contract, configured to atomically mark both a zero-knowledge proof digest and the biometric commitment as consumed upon a first successful enrollment, and to reject any subsequent enrollment attempt presenting either value; and an enrollment smart contract configured to verify a signature from an authorized attestor over the commitment, the proof, and a confidence score bound to a requesting contract address and chain identifier, and, upon successful verification and registry consumption, to mint a non-transferable token to the requesting wallet, the token's transfer and approval functions being disabled by construction.