ARTFEED — Contemporary Art Intelligence

Soft Redaction of Image Provenance via Zero-Knowledge Proofs

ai-technology · 2026-08-10

A new research paper on arXiv (2608.07063) proposes 'soft redaction' for image provenance, using zero-knowledge proofs (ZKPs) to replace sensitive provenance assertions with proofs of selected properties over hidden data. The work addresses the conflict between provenance transparency (e.g., C2PA standards) and privacy, where signed records of origin and editing history may reveal sensitive information. The authors demonstrate distance proofs within ZKP circuits, first for location assertions using Chebyshev polynomial approximations, then for L2 distance proofs over biometric embeddings to support privacy-preserving likeness claims. The approach aims to enforce personality rights while maintaining trust in digital images. The paper is categorized as a cross-type announcement and is available at arxiv.org/abs/2608.07063.

Key facts

  • Paper arXiv:2608.07063 proposes soft redaction for image provenance.
  • Uses zero-knowledge proofs (ZKPs) to replace sensitive provenance assertions.
  • Focuses on distance proofs within ZKP circuits.
  • Location assertions support proofs of proximity using Chebyshev polynomial approximations.
  • Extends to L2 distance proofs over biometric embeddings.
  • Aims to enforce personality rights with images.
  • Addresses conflict between provenance transparency and privacy.
  • Published on arXiv with announcement type 'cross'.

Entities

Institutions

  • arXiv

Sources