Live Indexing Active

The Open Ledger of Intent

We maintain a permanent, immutable record of public commitments, model cards, and system cards made by AI developers. Ensuring promises aren't silently scrubbed from history.

Explore the Archives Learn How It Works
4,291 Snapshots Archived 83 Labs Monitored 12 Policy Retractions Detected (Last 30d)

Our Core Values

Designed for journalists, researchers, and history of science scholars, the Registry operates on three fundamental pillars to ensure total historical accountability.

Archive Library

Truth

We verify and timestamp every document. Once entered into the ledger, a record cannot be altered without a visible audit trail.

Archival Integrity

Utilizing semantic data structures and decentralized storage principles to prevent link rot and revisionist history.

Open Access

Data is public infrastructure. All snapshots, diffs, and shadow cards are available via API and bulk download for archivists and librarians.

For Researchers

Track the evolution of safety claims over time. Compare "Shadow Cards" against official releases.

Access API Docs →

Tools for Accountability

We provide specialized tools to parse the legal and technical obfuscation often found in AI documentation.

diff_viewer.exe

The Promise Tracker

A "Wayback Machine" specifically tuned for Terms of Service, Acceptable Use Policies, and Safety Guidelines. We highlight silent deletions of clauses regarding military use, data privacy, and liability.

- "We will not use customer data for training."
+ "We may use customer data to improve services."
model_card_v2.md

Shadow Model Cards

Community-maintained documentation that fills in the gaps left by official, redacted model cards. Inspired by EU AI Act transparency requirements, we document what the labs won't.

  • Training Data Provenance
  • Unfiltered Bias Benchmarks
  • Compute Usage Estimates

Recent Ledger Entries

Real-time updates to the registry.

Timestamp Entity Document Type Change Type Status
2023-10-24 14:02 OpenAI Model Card (GPT-4) Redaction Detected Verified
2023-10-23 09:15 Anthropic Terms of Service Policy Update Verified
2023-10-22 18:45 Meta AI Llama 3 System Card New Entry Pending Review
2023-10-21 11:30 Google DeepMind Safety Filter Config Silent Removal Verified

Contribute to the Archive

This project relies on data librarians, open-source developers, and whistleblowers. Help us build the semantic history of AI development.