Show HN: Aicpm – Verifiable AI provenance labels for web content
1 points
1 hour ago
| 0 comments
| HN
AICPM (Authenticated AI Content Provenance Marking) is an open, provable way to label AI-generated text inside documents using signed chunk-level provenance.

Many current approaches (watermarks, detectors, platform labels) are brittle, unverifiable, or proprietary. AICPM uses provider-signed manifests and offline verifiers to produce transparency without central control or guessing.

Features: - Provider-signed chunk provenance with signatures - Deterministic verification (offline, browser-native) - Test vectors + reference verifier included - Browser extension shows “AI %”, verified/edited breakdown - Editor demo + mock provider for experimentation

Repo: https://github.com/Chattadude/aicpm Demo: (optional link to your demo video if you make one)

This is a reference architecture, not a detector or policy tool. It’s designed to support tool and platform adoption with clear verification semantics.

Happy to answer questions or clarify design tradeoffs.

No one has commented on this post.