What are the EU AI Act penalties and fines?

The EU AI Act's penalty regime, set out in Article 99 of Regulation (EU) 2024/1689, creates three tiers of administrative fines that scale with the severity of the breach, from a lowest ceiling of EUR 7 500 000 up to a top ceiling of EUR 35 000 000 or a percentage of worldwide turnover, whichever figure is higher.

The penalties chapter has applied since a fixed date in 2025, while a fine for any specific obligation only becomes possible once that obligation itself is in force under the Act's wider implementation timetable.

What are the three penalty tiers under Article 99?

Breaching the Article 5 ban on prohibited AI practices sits in the top tier: administrative fines of up to EUR 35 000 000 or, if the offender is an undertaking, up to 7 % of its total worldwide annual turnover for the preceding financial year, whichever is higher.

Most other operator duties sit in the middle tier, covering providers, authorised representatives, importers, distributors, deployers, notified bodies and the Article 50 transparency rules, and carry fines of up to EUR 15 000 000 or, if the offender is an undertaking, up to 3 % of its total worldwide annual turnover for the preceding financial year, whichever is higher.

The lowest tier applies to a narrower failure, the supply of incorrect, incomplete or misleading information to notified bodies or national competent authorities in reply to a request, shall be subject to administrative fines of up to EUR 7 500 000, or, if the offender is an undertaking, up to 1 % of its total worldwide annual turnover for the preceding financial year, whichever is higher.

Who enforces the fines, and does it depend on the system?

National market surveillance authorities remain the default supervisor for everything not reallocated elsewhere, high-risk systems, the Article 5 prohibitions, and the Article 50 transparency duties, in each Member State where the system is on the market.

The AI Office instead has exclusive supervision of AI systems built on a general-purpose AI model where the model and the system come from the same provider, and of AI systems integrated into very large online platforms and search engines under the DSA.

It is also the sole enforcer against general-purpose AI model providers.

For providers of general-purpose AI models specifically, enforcement sits not with the member states but with the European Commission, in practice the AI Office, and Article 101 gives the Commission the power to fine those providers up to 3% of worldwide annual turnover or 15 million euros, whichever is higher.

That Article 101 ceiling is notably lower than the Article 99(3) tier, which surprises people who assume foundation model providers face the largest exposure.

A separate track covers fines against Union institutions, bodies, offices and agencies that fall within the scope of the Regulation, handled outside the Article 99 structure.

What is a periodic penalty payment, and how is it different from a fine?

Beyond the one-off fines in Article 99, Article 75c lets the AI Office impose periodic penalty payments of up to 5% of average daily income or worldwide annual turnover in the preceding financial year, per day, for as long as a breach continues.

Because that charge accrues every day rather than being capped once, a sustained breach can in principle run past the headline one-off ceilings over time. Article 75c also sets a five-year limitation period for the AI Office both to exercise its powers and to enforce its decisions.

How do authorities decide how large a fine to impose?

Article 99(7) lists circumstances authorities must weigh, starting with the nature, gravity and duration of the infringement and of its consequences, taking into account the purpose of the AI system, as well as, where appropriate, the number of affected persons and the level of damage suffered by them.

Authorities also check whether administrative fines have already been applied by other market surveillance authorities to the same operator for the same infringement, and whether administrative fines have already been applied by other authorities to the same operator for infringements of other Union or national law.

Two factors sit largely in an organization's own hands after a breach occurs: the degree of cooperation with the national competent authorities, in order to remedy the infringement and mitigate the possible adverse effects of the infringement, and any action taken by the operator to mitigate the harm suffered by the affected persons.

How a breach came to light matters too, since the manner in which the infringement became known to the national competent authorities, in particular whether, and if so to what extent, the operator notified the infringement, is itself one of the listed factors.

Taken together, self-notification, cooperation and mitigation are three of the nine factors that weigh on the eventual amount.

Are fines lower for small and medium-sized companies?

Yes. For SMEs including start-ups, Article 99(6) inverts the general rule: each fine is up to the percentage or the fixed amount, whichever is lower.

The regulation text confirms the reversal directly: each fine referred to in this Article shall be up to the percentages or amount referred to in paragraphs 3, 4 and 5, whichever thereof is lower.

Without that inversion, a small operator committing a high-risk breach would face the full fixed sum regardless of size: without the inversion, a €4 million start-up committing a high-risk breach would face the €15 million fixed sum, roughly four times its annual revenue, and a straightforwardly terminal outcome.

A later amendment extended comparable treatment to small mid-cap enterprises for the Article 99(4) and (5) tiers.

When do these penalties actually start applying?

A fine can only follow a breach of an obligation that is already in force, so the penalties chapter has formally applied since 2 August 2025, with the exception of Article 101.

The prohibition of certain AI practices has applied since February 2025 and is therefore the first real ground for fines, immediately in the heaviest tier.

The GPAI obligations have applied since August 2025.

The transparency obligations of Article 50 and the AI literacy duty have been enforceable since August 2026.

The high-risk obligations, including the deployer obligations of Article 26, follow after the Digital Omnibus on 2 December 2027 for Annex III systems and 2 August 2028 for Annex I systems.