EU AI Act Compliance Self-Audit
Walk through a structured, 5-step audit to assess your company's AI systems against the EU AI Act. Identify risk classifications, documentation gaps, and build your compliance roadmap.
AI System Inventory
Risk Classification
Documentation Gaps
FRIA Assessment
Compliance Report
Free Self-Audit
$99 /mo
For ongoing monitoring, alerts, and audit trail
1. AI System Inventory
List every AI system your organization uses, deploys, or develops. This includes third-party tools with AI features (e.g., chatbots, hiring screeners, recommendation engines).
2. Risk Classification Details
For systems flagged as High Risk or above, answer these detailed questions to confirm classification. Systems classified as Minimal or Limited Risk are listed at the bottom.
EU AI Act Risk Tiers:
Unacceptable Risk (Art. 5) -- Banned. Social scoring by governments, real-time biometric ID in public spaces, manipulative/exploitative AI.
High Risk (Art. 6, Annex III) -- Permitted with strict obligations: conformity assessment, CE marking, registration, ongoing monitoring.
Limited Risk (Art. 50) -- Transparency obligations: users must know they are interacting with AI.
Minimal Risk -- No specific obligations beyond voluntary codes of conduct.
Unacceptable Risk (Art. 5) -- Banned. Social scoring by governments, real-time biometric ID in public spaces, manipulative/exploitative AI.
High Risk (Art. 6, Annex III) -- Permitted with strict obligations: conformity assessment, CE marking, registration, ongoing monitoring.
Limited Risk (Art. 50) -- Transparency obligations: users must know they are interacting with AI.
Minimal Risk -- No specific obligations beyond voluntary codes of conduct.
3. Documentation Requirements (Annex IV)
For each High-Risk AI system, the EU AI Act requires specific technical documentation. Check items your organization has already completed.
4. Fundamental Rights Impact Assessment (FRIA)
Article 27 requires deployers of high-risk AI systems to conduct a Fundamental Rights Impact Assessment before deployment. Rate the potential impact of each system on these fundamental rights.
Why FRIA matters: The FRIA is mandatory for deployers of high-risk AI systems in public-sector contexts and recommended for private deployers. It documents potential adverse impacts on fundamental rights and mitigation measures. Failing to conduct one may result in fines up to 1.5% of global annual turnover.
EU AI Act Compliance Self-Audit Report
Generated by CarbonHelm on
5. Compliance Report
Summary of your EU AI Act compliance status with actionable next steps.