Enforcement
EU AI Act fines and penalties
The AI Act sets the highest maximum fines in EU digital regulation history: up to €35M or 7% of global turnover. Here is exactly how the three fine tiers work.
The three tiers
How much can you actually be fined?
Prohibited practices
Article 5 violations
€35,000,000
or 7% of global turnover, whichever is higher
Deploying a banned practice: social scoring, real-time public biometric identification for law enforcement, manipulative or exploitative systems.
High-risk system breaches
Chapter III violations
€15,000,000
or 3% of global turnover, whichever is higher
Missing technical documentation, no conformity assessment, no human oversight, or failing to register a high-risk system in the EU database.
Incorrect information
Article 99 violations
€7,500,000
or 1% of global turnover, whichever is higher
Supplying incorrect, incomplete, or misleading information to notified bodies or national authorities during a compliance check.
Enforcement
Who enforces AI Act fines?
National market-surveillance authorities
Each EU member state designates an authority responsible for investigating and fining non-compliant AI systems in its territory, similar to how GPSR enforcement works country by country.
The EU AI Office
Coordinates enforcement for general-purpose AI models at EU level and can investigate providers of the largest, most capable models directly.
Market withdrawal powers
Authorities can order a system pulled from the EU market independent of any fine. For a product-led company, losing EU market access can cost more than the fine itself.
Context
How AI Act fines compare to other EU laws
Related guides
Complete EU AI Act guide
The 4 risk tiers, obligations, deadlines and fines in one place
Read guide →
AI Act risk levels explained
Prohibited, high, limited and minimal risk, with real examples
Read guide →
Is your chatbot high-risk?
What Article 50 requires from customer service bots and assistants
Read guide →