What is the HUMMBL EU AI Act readiness audit?

The HUMMBL EU AI Act readiness audit is a free 30-minute assessment that evaluates whether your AI agent system meets Annex III requirements. It covers technical documentation, human oversight, accuracy and robustness, and cybersecurity obligations.

Does Annex III of the EU AI Act apply to my system?

Annex III applies to AI systems in 8 high-risk categories: biometric identification, critical infrastructure, education, employment, essential services, law enforcement, migration management, and administration of justice. The HUMMBL audit includes a checklist to determine whether Annex III applies to your specific deployment.

When does EU AI Act enforcement begin?

EU AI Act enforcement for high-risk AI systems under Annex III is expected to apply from December 2, 2027, per the May 2026 provisional agreement on the Digital Omnibus. This date is subject to formal adoption by the Council and Parliament. Organizations deploying high-risk AI systems in the EU should prepare conformity assessments, technical documentation, and human oversight mechanisms accordingly.

HUMMBL Audit

Is Your Agent System
EU AI Act Ready?

Annex III enforcement expected December 2, 2027 (per May 2026 provisional agreement). High-risk AI systems must demonstrate compliance. Here's what your agents need.

days until EU AI Act Annex III enforcement target. Penalties vary by violation category: up to €35M or 7% for prohibited practices (Article 5), up to €15M or 3% for high-risk non-compliance (Article 99), whichever is higher.

€15M

Max Fine (High-Risk)

Of Global Turnover

Required Systems

What's Required

Annex III Requirements

Risk Management System

Continuous risk identification, analysis, and mitigation across your AI agent's lifecycle.

Article 9

Automatic Logging

Traceable, auditable logs of every agent decision and action.

Article 12

Human Oversight Measures

Humans can understand, monitor, and override agent behavior at any time.

Article 14

Accuracy & Robustness

Documented performance levels, error rates, and adversarial resilience.

Article 15

Data Governance

Training data quality standards, bias monitoring, and lifecycle documentation.

Article 10

Technical Documentation

Complete system description for regulatory conformity assessment.

Article 11

Self-Assessment

Readiness Checklist

Agents produce append-only audit logs of all actions
Governance traces map to regulatory controls
Kill switch can halt agent execution immediately
Delegation chains are signed and depth-limited
Humans can review and override any agent decision
Agent identity is cryptographically verifiable
Compliance evidence bundles generated on demand
Risk management process is documented and versioned
Performance metrics and error rates are tracked
Data processing records exist per GDPR Art. 30

Affected Domains

Does Annex III Apply to You?

If your AI agents operate in any of these regulated domains, Annex III high-risk classification applies:

Financial services

Healthcare

Employment & HR

Education

Law enforcement

Critical infrastructure

Immigration & asylum

Democratic processes

15,600+ platform tests in production

0 third-party dependencies

hummbl-governance on PyPI

Apache 2.0 open-source license

Get Your Free Assessment

30-minute call. I'll review your agent architecture and tell you exactly where your compliance gaps are. No commitment required.

Book a Free Assessment reuben@hummbl.io

Boundary Disclaimer

—Enforcement dates are per the May 2026 provisional agreement on the Digital Omnibus and are subject to formal adoption by the EU Council and Parliament.
—Penalty tiers shown are per Article 99 of the AI Act. The €35M/7% maximum applies to prohibited practices (Article 5). High-risk non-compliance falls under Tier 2 (€15M/3%).
—HUMMBL is not a Notified Body under the EU AI Act. This assessment maps technical readiness; it does not constitute a legal conformity assessment.
—For authoritative regulatory guidance, consult the EU AI Act Explorer maintained by the Future of Life Institute.