Skip to main content

Universal Content Fingerprinting Registry (UCFR)

A comprehensive, privacy‑first way for creators to establish verifiable authorship with cryptographic fingerprints and an on‑chain registry, while transparently declaring AI involvement.

Why UCFR?

  • Simplified model: Three core structs — Claim, ExternalID, Method.
  • Neutral core: No policy thresholds on‑chain; applications apply their own logic.
  • Future‑proof: Methods are islands; new methods add new IDs without migrations.
  • Open‑source first: Built on widely available, community‑verified methods; vendor‑neutral.
  • Composable: Works with C2PA and W3C VCs; on‑chain metadata uses canonical JSON (RFC 8785).

Who is involved?

NIST
ISO
OpenAI
Meta
Google
Microsoft
OpenCV
Shazam

Logos shown for context; UCFR is a neutral public good and not affiliated with these organizations.

Claim

  • fingerprint: bytes
  • methodId: uint16
  • externalId: uint16 (optional)
  • externalSig / pubKey: bytes (optional)
  • metadata / extURI: string (optional)
  • owner, timestamp

Keyed by digest of (fingerprint, methodId, extId). Immutable once written.

ExternalID

  • extId: uint16
  • specURI: string
  • sigSizeHint: uint32 (0 = variable)
  • isSymmetric: bool
  • active: bool

Signature/MAC schemes (e.g., RSA‑PSS, ECDSA, Ed25519, HMAC).

Method

  • methodId: uint16
  • name: string
  • specURI: string
  • fpSizeBytes: uint32
  • active: bool

Hashing algorithms (e.g., SHA‑256, BLAKE3, PDQ).

Desktop Application

Real‑time folder monitoring, automatic hash generation and signing, metadata with AI declarations, and a simple UI that hides blockchain complexity.

GeneralizedClaimRegistry

On‑chain registry for cryptographic claims with multiple hashing methods, optional external signatures/MACs, immutable storage, and batch submission.

Blockchain Infrastructure

Built on the Stability network with ZKT support for streamlined transactions and public, immutable verification.

Workflow

  1. Configure defaults and select folders to monitor.
  2. App detects new files and computes fingerprints (SHA‑256 by default).
  3. Checks the registry for existing claims.
  4. New files are signed and submitted on‑chain.
  5. Existing claims are compared; mismatches prompt review.
  6. Visual status keeps track of verification and submission state.

What makes UCFR different?

  • Hijack‑resistant: Perceptual hashes can't claim ownership; only exact digests can.
  • Simple and stable: Compact on‑chain core; simple to integrate.
  • Interoperable: Designed to snap into C2PA, VC 2.0, and common fingerprint libraries.

Method examples by media type

Image

  • PDQ, pHash, aHash, dHash, OpenCV BlockMeanHash, ColorMomentHash

Audio

  • Chromaprint/AcoustID, Shazam‑style, audfprint

Video

  • TMK+PDQF, vPDQ, OpenCV frame hashes

File/Generic

  • SHA‑256, SHA3‑256, BLAKE3

Text

  • MinHash, SimHash, Winnowing

Device/Camera

  • PRNU, Noiseprint

These are open‑source friendly examples; UCFR remains extensible without migrations (island model).

Security & Privacy

  • Local key storage: Private keys are generated, encrypted, and stored locally. They never leave your device.
  • Transparent AI declarations: Per‑file metadata can declare AI involvement for downstream transparency.
  • Decentralized verification: Anyone can verify claims on‑chain without trusting a central authority.
  • Optional crash reporting: Privacy‑preserving diagnostics with no keys or sensitive contents.

Immutable by design

Claims are append‑only and cannot be altered or deleted, providing durable provenance records.