Compliance

Turn AI governance into evidence you can hand over

MERIDVAR helps organisations operationalise governance controls and generate evidence relevant to the frameworks they answer to. Compliance itself remains your determination. MERIDVAR gives you the record to support it.

  • Mapped: EU AI Act, NIST AI RMF and ISO/IEC 42001
  • Control status computed from the running appliance
  • MERIDVAR issues no certification
Control evidenceIllustrative

ISO/IEC 42001, Clause 8: operational controls Partial

Three of four observations are satisfied.

ObservationMeasuredResult
Prompt DLP blocking categories6Satisfied
Threat detection enforcing on live trafficYesSatisfied
Teams with an enforced residency policy0 of 4Not satisfied
Vendor risk requirements enforced at the callYesSatisfied

A control that the product cannot observe is reported as uncomputed, with the reason.

Computed status

A control status that reads the running system

The status of each mapped control is derived from named observations of the live appliance. Each observation shows what was measured and whether it satisfied the control. Switch off a control, and the status changes.

  1. AI systemsThe inventory, with each system’s governance state
  2. ControlsThe policy enforced at the gateway
  3. EvidenceObservations computed from the appliance, and the evidence log
  4. FrameworkA mapped control and its computed status
  5. AuditA package for each AI system and each vendor

By territory

What is mapped, and what is supporting evidence

There are two categories. Mapped means that the product ships a control mapping for the framework. Evidence means that records in MERIDVAR are relevant to the framework, with no mapping in the product.

FrameworkStatusWhat MERIDVAR contributes
Europe
EU AI ActMappedA control mapping with computed status. The risk rules read the organisation’s own EU AI Act classification of each system.
GDPREvidencePrompt DLP on personal identifiers, a data residency policy and a privacy review that records the organisation’s DPIA position.
DORAEvidenceVendor due diligence, vendor approval and the register of AI systems that depend on each provider.
NIS2EvidenceAccess roles, the audit trail and security review records.
United Kingdom
UK GDPR and the Data Protection Act 2018EvidenceThe same privacy review, DLP and residency evidence as for the GDPR.
Regulator-led AI principlesEvidenceInventory, ownership, risk classification and approval history for each AI system.
United States
NIST AI RMFMappedGovern, Map, Measure and Manage, each with observations computed from the running appliance.
State AI and privacy lawsEvidenceRecords of systems whose outputs support decisions about individuals, and of human review.
International standards
ISO/IEC 42001MappedClauses 5 to 9 and the supplier controls of Annex A, with computed status.
ISO/IEC 27001 and SOC 2EvidenceAccess control and logging evidence for your own programme.
Latin America
LGPD (Brazil)EvidencePrompt DLP on personal identifiers, a residency policy and privacy review records.
Asia Pacific and the Middle East
National AI governance frameworks and data protection lawsEvidenceThe governance record, the residency policy and the evidence log apply regardless of framework. No regional framework is mapped in the product today.

Framework summaries are drawn from public sources and are not legal advice. Your own counsel should confirm what applies to you.

EU AI Act timeline

Scoped to the obligations that are actually on the calendar

The Digital Omnibus on AI moved the high-risk deadlines. The transparency and general-purpose model obligations did not move. MERIDVAR’s mapping reflects the amended timeline.

  1. 2 February 2025Prohibited practicesApplies
  2. 2 August 2025General-purpose AI model obligationsApplies
  3. 2 August 2026Transparency obligations under Article 50Applies
  4. 2 December 2027High-risk systems listed in Annex IIIAhead
  5. 2 August 2028High-risk systems embedded in products under Annex IAhead
Audit packageIllustrative

Support reply assistant

One downloadable package for each AI system and each vendor.

Governance stateIncluded
Classifications, reviews and approvalsIncluded
Assessments, with the questionnaire versions answeredIncluded
Evidence records, with hash, previous hash and signatureIncluded
Public signing keysIncluded

The package states what it does not prove.

The audit package

Evidence that is explicit about its own scope

What each package holds

  • The subject’s governance state
  • Its classifications, reviews and approvals
  • Its assessments, with the exact questionnaire versions answered
  • Every evidence record about it, as stored, with its hash, the previous hash and its signature
  • The public signing keys

What it states about its scope

  • Each record can be checked against its hash and signature
  • Signatures are made with a key held on your appliance
  • Statements are recorded as stated, with who made them and when

The product views on this page are illustrative, with invented example data.

See MERIDVAR refuse a request on your own network

A working session with the founder, on a host you control.

What happens next

  1. You send the formYour request goes straight to the founder.
  2. We reply by emailWe agree a time and what you want to see.
  3. We run a working sessionWe install the appliance with you, route a test request through it and read the evidence it leaves.