Skip to main content

The Integrity Framework · v1.0 · Crosswalks

How this framework maps to the standards you already cite.

The Integrity Framework is not trying to replace NIST AI RMF, ISO 42001, EU AI Act conformity, CSA AICM, SOC 2, or WCAG. It sits in the gap they don't fill: integrity for AI-powered SaaS that isn't itself a compliance product. The crosswalks below let you cite the framework alongside whatever regulatory regime your buyer recognizes.

Frozen v1.0·Last updated 2026-04-25·Cite as https://claritylift.ai/framework/v1/crosswalks

How to read this

Each row below maps a framework layer item to the closest equivalent control / requirement in each external regime. “Closest equivalent” is interpretive: most of these regimes have controls that span multiple framework items, and vice versa. Use these as starting points for procurement and audit conversations, not as substitutes for legal review.

N/A means the regime does not address the failure mode this framework item targets. Partial means the regime addresses part of the failure mode but not all of it. Full means a customer can rely on the regime alone for that item.

Layer 1 vetoes (pre-build)

Framework vetoNIST AI RMFISO/IEC 42001EU AI ActCSA AICMSOC 2 TSCWCAG
Veto 1: Artifact vs outcomeGOVERN-1.1 (mission)Clause 4.3 scopeN/A (regulates products, not pricing models)GOV-01 (governance framework)CC1.1 (entity values)N/A
Veto 2: IndependenceGOVERN-2.3 (accountability)Clause 5.3 rolesArticle 17 (post-market monitoring)AIM-01 (independent oversight)CC2.1 (independence)N/A
Veto 3: VerifiabilityMEASURE-2 (test, evaluate)Clause 9.1 monitoringArticle 15 (accuracy, robustness)AIS-04 (testing)CC7.1 (objective verification)4.1.2 (name, role, value)
Veto 4: AI accountabilityGOVERN-1.5 (oversight)Clause 7.2 (competence + review)Article 14 (human oversight)AIM-04 (human-in-the-loop)CC5.1 (control activities)N/A
Veto 5: Pricing-rigor alignmentGOVERN-3.1 (compliance-monitoring incentives)Clause 4.2 (interested-party expectations)N/AGOV-04 (incentive structures)CC2.2 (commitment to integrity)N/A
Veto 6: TechCrunch testMAP-2 (use-case context)Annex A.5 (impact assessment)Article 9 (risk management system)GOV-03 (risk management)CC3.1 (risk identification)N/A

Layer 2 architectural constraints

Framework constraintNIST AI RMFISO/IEC 42001EU AI ActCSA AICMSOC 2 TSCWCAG
Evidence chain integrityMEASURE-3 (documentation)Clause 7.5 (documented information)Article 12 (record-keeping)AIS-02 (data lineage)CC7.2 (system monitoring)N/A
AI output review gatesMANAGE-2.3 (response + recovery)Annex A.7 (human oversight)Article 14 (human oversight)AIM-04 (human-in-the-loop)CC8.1 (change management)4.1.3 (status messages)
Customer self-attestation isolationMAP-3 (data documentation)Annex A.6 (data quality)Article 10 (data governance)AIS-03 (data classification)CC6.1 (logical access)3.3.1 (error identification)
ReproducibilityMEASURE-2.5 (validity)Clause 9.1 monitoringArticle 15 (accuracy)AIS-04 (testing)CC7.1 (objective verification)N/A
Evidence retention independenceMANAGE-1.2 (incident response)Annex A.8 (records management)Article 12 (record-keeping)GOV-02 (records management)A1.2 (availability)N/A
Independent verification hooksGOVERN-5.1 (audit)Clause 9.2 internal auditArticle 17 (post-market monitoring)GOV-05 (independent audit)CC4.1 (monitoring activities)5.1 conformance claims
Failure transparencyMEASURE-2.7 (impact assessment)Annex A.10 (incident reporting)Article 62 (serious-incident reporting)AIS-05 (incident handling)CC2.3 (information communication)3.3.3 (error suggestion)

Layer 3 operational guardrails

Framework guardrailNIST AI RMFISO/IEC 42001EU AI ActCSA AICMSOC 2 TSCWCAG
Refund-on-failure clauseN/AClause 8.3 (operational planning)N/A (contractual)GOV-04 (commercial alignment)CC2.2 (commitment to integrity)N/A
Public methodology pageGOVERN-1.6 (transparency)Annex A.4 (transparency)Article 13 (transparency to deployers)AIM-03 (disclosure)CC2.3 (communication)5.1 conformance claims
Annual independent auditGOVERN-5.1Clause 9.2 internal + 9.3 management reviewArticle 17 (post-market)GOV-05 (independent audit)CC4.1 (monitoring)N/A
Customer-side compliance ownerGOVERN-2.1 (roles)Clause 5.3 (org roles)Article 16 (deployer obligations)AIM-02 (accountability)CC1.4 (commitment to competence)N/A
Whistleblower channelMANAGE-4.3 (incident reporting)Annex A.10 (incident reporting)Article 62 (serious-incident reporting)AIS-05 (incident handling)CC2.3 (communication)N/A
Accountability communityGOVERN-1.6 (transparency to stakeholders)Clause 4.2 (interested parties)N/AAIM-03 (disclosure)CC2.3 (communication)N/A
Public kill criteriaMANAGE-2.4 (decommission)Annex A.9 (lifecycle management)Article 17 (post-market)AIS-06 (decommissioning)CC8.1 (change management)N/A

What the crosswalk doesn't tell you

The framework sits in a gap that all six regimes leave open: integrity for AI-powered SaaS that isn't itself a compliance product. NIST AI RMF and ISO 42001 govern AI systems where the AI is the product. EU AI Act governs high-risk AI products. CSA AICM is a control matrix designed for institutional adoption. SOC 2 covers organizational security. WCAG covers accessibility. None of them cover the case where AI is a feature inside a product whose primary value is something else — operations tooling, decision intelligence, accessibility documentation, organizational health.

That is the gap the framework fills. Crosswalks are how the framework borrows credibility from regimes the buyer already recognizes; they are not a replacement for the framework's own substance.