MECHASSURE-IEB / Version 0.1Open Standard · Free Verifier
MECHASSURE.org
Machine Evidence · Coverage Handoff · Assurance Standard
MECHASSURE-IEB / 0.1
Public Draft · 2026-06

MECHASSURE Insurer Evidence Bundle Profile

A MECHASSURE Insurer Evidence Bundle is a BAINK-compatible evidence bundle that contains enough preserved evidence for an insurer or reinsurer to evaluate AI-assisted legal, operational, financial, technical, or compliance decisions.

Standard metadata
001Profile: mechassure-ieb-0.1
002Status: Public Draft
003Base profile: legal-oreo-lr-0.1
004Family: BAINK · Legal OREO · MECHASSURE
005License: Open standard / free verifier profile
006Editors: MECHASSURE working group
§ 1

Core principle

Insurers should not have to trust an AI vendor's narrative. They should be able to verify the record. MECHASSURE describes the minimum file set, the hash discipline, and the insurer-facing maps required to make that verification deterministic.

§ 2

Conformance

A bundle conforms to mechassure-ieb-0.1 when:

  1. It is a valid BAINK bundle (manifest, hash chain, receipt).
  2. It embeds a valid legal-oreo-lr-0.1 sub-bundle.
  3. It includes insurance/coverage_map.json, insurance/exclusions_map.json, and insurance/risk_controls.json.
  4. It emits a reports/mechassure_report.json from the public verifier.
  5. Every artifact in the manifest is hash-matched at validation time.
§ 3

Required artifacts

bundle.profile.artifacts
001[
002 { "path": "legal-oreo-bundle.json", "type": "legal_oreo_bundle", "required": true },
003 { "path": "insurance/coverage_map.json", "type": "coverage_map", "required": true },
004 { "path": "insurance/exclusions_map.json", "type": "exclusions_map", "required": true },
005 { "path": "insurance/risk_controls.json", "type": "risk_control_map", "required": true },
006 { "path": "reports/mechassure_report.json","type": "insurer_verification_report","required": true }
007]
§ 4

Non-claims

The profile makes no statement about:

  • The correctness of the underlying legal conclusion.
  • Whether the insurer should issue, modify, or deny coverage.
  • Whether an AI model is safe, accurate, or fit for purpose.
§ 5

Related profiles