ISO 42001Clauses 6.1, 8.2–8.4, 9.1
AI management systems
Model cards, fairness and bias documentation, and judge audit events map to the AIMS controls that ask you to evidence ongoing AI risk management.
AI compliance
Evidence tied to traffic, not assertions in a questionnaire.
Governance frameworks ask you to show that a control ran, on what, and with what result. ZNYX produces those artifacts as a by-product of enforcement, generated per environment from the policy that actually executed.
EU AI ActArt. 12, 13, 14, 15
High-risk obligations
Logging, human oversight hooks, accuracy and robustness evidence, and technical documentation for systems placed on the EU market.
NIST AI RMFMEASURE 2.x, MANAGE 2.x
Govern · Map · Measure · Manage
Detector scorecards and drift monitoring provide the Measure function; policy versioning and rollout provide Manage.
OWASP LLMLLM01–LLM10
Top 10 for LLM Applications
A per-environment coverage scorecard generated from the detectors your policy actually enables.
Workflow
From policy to audit answer
01
Declare the policy
Pin a policy bundle version per environment so the control you evidence is the control that ran.
02
Enforce and record
Every decision writes an append-only audit event with trace id, detector verdicts, and policy version.
03
Generate evidence
Export the coverage scorecard, detector metrics, and model cards for the period under review.
04
Answer the reviewer
Hand over artifacts tied to real traffic instead of a questionnaire full of assertions.
Scope of this claim
ZNYX produces evidence that supports these frameworks. It does not certify you against them, and ZNYX itself does not currently hold SOC 2 or ISO 27001 attestation, those are on the roadmap and are not claimed today. Mapping to specific clauses is provided to accelerate your own assessment, not to replace it.
FAQ
AI compliance questions
What evidence ZNYX produces, which frameworks it maps to, and where certification stops.
No, and no tool can. ZNYX produces the evidence that supports an assessment: per-decision audit records, OWASP LLM Top-10 coverage scorecards, detector scorecards with measured precision and recall, and ISO 42001-style model cards. Mapping to specific clauses is provided to accelerate your own assessment, not to replace it.
A coverage scorecard scored 0-100 against the 2025 OWASP Top 10 for LLM Applications, detector scorecards recording how each detector was measured, model cards for the classifiers and judges in the path, benchmark runs against your own datasets, and an audit trail of policy versions with the reason each change was made.
The EU AI Act, ISO/IEC 42001, and the NIST AI Risk Management Framework, with the OWASP LLM Top 10 as the technical control mapping underneath them.
Not today. ZNYX does not currently hold SOC 2 or ISO 27001 attestation; those are on the roadmap and are not claimed. Security and performance claims on this site reflect implemented capabilities and internal benchmarks rather than external audit.