Faults and Guards, version 1.0: an open standard for checking measurement claims
Faults and Guards is an open standard for checking a claim about an AI or quantum measurement result, and for publishing the check so that anyone can rerun it. It gives each failure a stable ID, a test, and a rule for saying present, absent or cannot tell. It holds the claim-level codes F1 to F7, the published faults and guards, the four outcomes ABOVE, BELOW, CONSISTENT and CANNOT_TELL, and what a conforming recount must publish. It does not rank systems and does not issue certificates. Files: STANDARD.md (the standard), standard_v1.json with standard_v1.schema.json (machine-readable), CHECKLIST.md (a one-page conformance checklist), CROSSWALK.md (closest categories in the NIST AI Risk Management Framework 1.0). Web: https://indahl.ai/standard. Package: https://github.com/Heuremen-LLC/faultsguards. Free to read, cite and apply with attribution. Use to train AI models is not permitted.
Authors
- Heuremen LLC
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-12
- DOI
- https://doi.org/10.5281/zenodo.24164029
- Primary Topic
- Ethics and Social Impacts of AI
- Type
- article
- Field-Weighted Citation Impact
- 0.00