The EU is tightening its AI regulation: Starting August 2026, providers of high-risk AI systems must demonstrate full compliance with the EU AI Act. Violations carry penalties of up to 7% of global annual revenue.
High-Risk Systems in Focus
The EU AI Act categorizes AI systems by risk class. Particularly affected are:
- Human resources: CV screening, applicant selection
- Credit sector: Credit scoring, loan decisions
- Law enforcement: Predictive policing, biometric ID
- Critical infrastructure: Transport, energy, water
7% Revenue Fine as Deterrent
The draconian penalties are modeled on the GDPR and can mean billions in losses for large corporations. A technology company with 100 billion in revenue risks a 7 billion euro fine.
Compliance Rush Until August
Companies have until August 2026 for:
- Risk assessment: Classification of all AI systems
- Documentation: Complete compliance records
- Testing: Bias tests and robustness validation
- Governance: Implementing AI governance structures
US technology corporations in particular must fundamentally overhaul their EU operations. Some are considering withdrawing from the EU market for certain AI services.
Sources: artificialintelligenceact.eu, kennedyslaw