# EU AI Act Penalties and Prohibited Practices: What Article 5 Bans, What Each Breach Can Cost, and Who Enforces It

Source: https://superintelligencecompliance.ai/guides/eu-ai-act-penalties-prohibited-practices  
Publisher: ZSKFL Management  
Last updated: 2026-11-02. Not legal advice.

> EU AI Act fines, whichever is higher: €35M or 7% for Article 5 bans, €15M or 3% for listed duties, €7.5M or 1% for incorrect or misleading information.

## At a glance

- EU AI Act fines reach €35 million or 7% of worldwide annual turnover, whichever is higher, for prohibited practices; €15 million or 3%, whichever is higher, for the operator and notified body duties listed in Article 99(4), including Article 50 transparency; and €7.5 million or 1%, whichever is higher, for supplying incorrect, incomplete, or misleading information to notified bodies or national competent authorities in reply to a request. For SMEs the lower figure applies (Article 99(6)).
- Article 5 bans eight practices and has applied since February 2, 2025. The Digital Omnibus on AI added two more, on non-consensual intimate material and child sexual abuse material, which apply from December 2, 2026.
- Member states lay down the penalty rules (Article 99(1)), and national market surveillance authorities enforce most of the Act. The Commission fines general-purpose AI model providers under Article 101 from August 2, 2026, and the European Data Protection Supervisor fines EU institutions under Article 100.
- Italy’s Legislative Decree No. 179 of October 7, 2026, in force since October 23, 2026, is one national law setting these fines.

## What are the maximum EU AI Act fines?

The maximum EU AI Act fine is €35 million or, if the offender is an undertaking, 7% of its total worldwide annual turnover for the preceding financial year, whichever is higher, for non-compliance with the Article 5 prohibitions (Article 99(3)). Breaches of the provider, authorised representative, importer, distributor, and deployer obligations in Articles 16, 22, 23, 24, 25(2) and (4), and 26, of the notified body requirements in Articles 31, 33(1), (3), and (4), and 34, and of the Article 50 transparency obligations carry up to €15 million or 3%, whichever is higher (Article 99(4)). Supplying incorrect, incomplete, or misleading information to notified bodies or national competent authorities in reply to a request carries up to €7.5 million or 1%, whichever is higher (Article 99(5)).

These are ceilings. For SMEs, including start-ups, each fine is capped at the lower of the two figures (Article 99(6)). The Digital Omnibus on AI, Regulation (EU) 2026/1744, extended the lower-figure rule to small mid-cap enterprises (SMCs), as defined in Commission Recommendation (EU) 2025/1099, but only for the €15 million and €7.5 million tiers (Article 99(6a)). An SMC that breaches Article 5 faces the same ceiling as any other undertaking: €35 million or 7% of total worldwide annual turnover, whichever is higher.

**EU AI Act fines by breach, Regulation (EU) 2024/1689 as consolidated on July 27, 2026**

| Breach | Maximum fine | SME rule | From when |
|---|---|---|---|
| Article 5 prohibited practices, points (a) to (h) (Article 99(3)) | €35 million or 7% of worldwide annual turnover, whichever is higher | Lower figure for SMEs (Article 99(6)); no SMC rule | Prohibitions since February 2, 2025; Article 99 since August 2, 2025 |
| Article 5 points (ba) and (bb), added by the Omnibus (Article 99(3)) | €35 million or 7%, whichever is higher | Lower figure for SMEs; no SMC rule | December 2, 2026 |
| Article 50 transparency obligations (Article 99(4)(g)) | €15 million or 3%, whichever is higher | Lower figure for SMEs and SMCs (Article 99(6) and (6a)) | August 2, 2026; December 2, 2026 for the Article 50(2) marking duty on generative systems placed on the market before August 2, 2026 |
| High-risk operator duties in Articles 16, 22, 23, 24, 25(2) and (4), and 26 (Article 99(4)) | €15 million or 3%, whichever is higher | Lower figure for SMEs and SMCs | December 2, 2027 for stand-alone Annex III systems; August 2, 2028 for AI in Annex I Section A regulated products |
| Notified body requirements in Articles 31, 33(1), (3), and (4), and 34 (Article 99(4)(f)) | €15 million or 3%, whichever is higher | Lower figure for SMEs and SMCs | August 2, 2025 |
| Incorrect, incomplete, or misleading information to notified bodies or national competent authorities (Article 99(5)) | €7.5 million or 1%, whichever is higher | Lower figure for SMEs and SMCs | August 2, 2025 |
| EU institutions, bodies, offices, and agencies (Article 100) | €1.5 million for Article 5; €750,000 for other requirements | Not applicable | Article 100 since August 2, 2025; each underlying requirement from its own date |
| General-purpose AI model providers, fined by the Commission (Article 101) | €15 million or 3%, whichever is higher | No SME rule in Article 101 | August 2, 2026 |
| AI systems within the AI Office’s exclusive competence under Article 75(1), fined by the AI Office (Article 75c(4)) | Article 99(3) to (7) applied mutatis mutandis: €35 million or 7% for Article 5; €15 million or 3% for any other applicable provision, including those not listed in Article 99(4); €7.5 million or 1% for incorrect, incomplete, or misleading information to the AI Office in reply to a request; whichever is higher in each case | Lower figure for SMEs; for SMCs in the €15 million and €7.5 million tiers only (Article 99(6) and (6a), applied mutatis mutandis) | August 2, 2026 for the fining power; each duty from its own date |

Chapter XII, which contains Articles 99 to 101, has applied since August 2, 2025, except Article 101, which applies from the general date of August 2, 2026 (Article 113(b)). The last column gives the date the duty applies, or the fine provision where that is later. For AI systems classified as high-risk under Article 6(1) that relate to Annex I Section B products, including machinery, only Article 6(1), Article 60a, and Articles 102 to 112 apply (Article 2(2)).

## How is the amount of a fine decided?

Article 99(7) requires the authority, when deciding whether to impose a fine and its amount, to take into account all relevant circumstances and, as appropriate, ten listed factors: the nature, gravity, and duration of the infringement and of its consequences, taking into account the purpose of the AI system and, where appropriate, the number of affected persons and the level of damage they suffered; whether other market surveillance authorities have already fined the same operator for the same infringement; whether other authorities have already fined the same operator for infringements of other Union or national law resulting from the same activity or omission; the operator’s size, annual turnover, and market share; any other aggravating or mitigating factor, such as financial benefits gained or losses avoided, directly or indirectly, from the infringement; the degree of cooperation with the national competent authorities to remedy the infringement and mitigate its possible adverse effects; the operator’s degree of responsibility, taking into account the technical and organisational measures it implemented; how the infringement became known to the national competent authorities, in particular whether, and if so to what extent, the operator notified it; whether the infringement was intentional or negligent; and any action the operator took to mitigate the harm suffered by the affected persons.

Article 99(1) also allows penalties to include warnings and non-monetary measures and, as amended by the Omnibus, requires member states to take into account the interests of SMEs, including start-ups, and SMCs, and their economic viability when imposing penalties. Each member state decides to what extent its public authorities can be fined (Article 99(8)).

## Which practices does Article 5 prohibit?

Article 5(1) prohibits placing on the market, putting into service, or using AI systems for the practices below. For points (d) to (g), the ban on putting into service covers putting the system into service “for this specific purpose,” and point (h) bans only the use.

**Article 5(1) prohibited practices, Regulation (EU) 2024/1689 as consolidated on July 27, 2026**

| Point | What is banned | Key exception or limit |
|---|---|---|
| (a) | Subliminal techniques beyond a person’s consciousness, or purposefully manipulative or deceptive techniques, with the objective or effect of materially distorting behaviour by appreciably impairing the ability to make an informed decision | Only where the resulting decision causes, or is reasonably likely to cause, significant harm to that person, another person, or a group |
| (b) | Exploiting vulnerabilities of a person or group due to age, disability, or a specific social or economic situation, with the objective or effect of materially distorting behaviour | Only where it causes, or is reasonably likely to cause, significant harm |
| (ba) | Generating or manipulating realistic images, videos, audio, or similar material of an identifiable natural person’s intimate parts, or of an identifiable person engaged in sexually explicit activities, without that person’s freely-given, specific, informed, unambiguous, and explicit consent for that generation or manipulation | Applies from December 2, 2026; providers and deployers caught only as set out in Article 5(1a); manipulation that does not increase exposure or alter the nature of the activity is excluded (Article 5(1b)) |
| (bb) | Generating or manipulating material or performance within the meaning of Article 2, points (c) and (e), of Directive 2011/93/EU, which define child pornography and pornographic performance | Applies from December 2, 2026; not where a “without right” defence applies under national law; same Article 5(1a) limits as (ba) |
| (c) | Evaluating or classifying people over time based on social behaviour or known, inferred, or predicted personal or personality characteristics, with the social score leading to detrimental or unfavourable treatment | Only where the treatment occurs in social contexts unrelated to where the data was generated or collected, or is unjustified or disproportionate to the behaviour or its gravity |
| (d) | Risk assessments of individuals to assess or predict the risk of committing a criminal offence, based solely on profiling or on assessing personality traits and characteristics | Not AI systems used to support the human assessment of a person’s involvement in a criminal activity, which is already based on objective and verifiable facts directly linked to a criminal activity |
| (e) | Creating or expanding facial recognition databases through the untargeted scraping of facial images from the internet or CCTV footage | No exception in the text |
| (f) | Inferring the emotions of a natural person in the areas of workplace and education institutions | Except where the system is intended to be put in place or into the market for medical or safety reasons |
| (g) | Biometric categorisation that categorises individuals based on biometric data to deduce or infer race, political opinions, trade union membership, religious or philosophical beliefs, sex life, or sexual orientation | Not labelling or filtering of lawfully acquired biometric datasets, such as images, or categorising biometric data in the area of law enforcement |
| (h) | Use of “real-time” remote biometric identification systems in publicly accessible spaces for law enforcement | Only where strictly necessary for one of three objectives, under the conditions in Article 5(2) to (7), and only where national law provides for it |

Point (f) is the ban most likely to reach an ordinary employer. The Commission’s guidelines read “workplace” to include candidates during selection and hiring, and give as prohibited examples emotion recognition during recruitment and a call centre using webcams and voice recognition to track its employees’ emotions; tracking customers’ emotions is not covered, and inferring emotions from written text is not based on biometric data, so it falls outside the ban (guidelines, paragraphs 251 and 254). Biometric and law enforcement uses that Article 5 does not ban can still be high-risk under Annex III, which the [high-risk classification guide](https://superintelligencecompliance.ai/guides/eu-ai-act-high-risk-ai-systems) covers.

## When may police use real-time remote biometric identification?

Only where strictly necessary for one of three objectives: the targeted search for specific victims of abduction, trafficking in human beings, or sexual exploitation, and the search for missing persons; the prevention of a specific, substantial, and imminent threat to life or physical safety, or a genuine and present or genuine and foreseeable threat of a terrorist attack; or the localisation or identification of a person suspected of having committed a criminal offence, for the purpose of conducting a criminal investigation or prosecution or executing a criminal penalty, for offences listed in Annex II and punishable in the member state concerned by a custodial sentence or detention order for a maximum period of at least four years (Article 5(1)(h)).

Each use may serve only to confirm the identity of the specifically targeted individual. The authority must have completed a fundamental rights impact assessment under Article 27 and registered the system under Article 49, although in duly justified cases of urgency use may begin before registration if registration follows without undue delay (Article 5(2)). Each use needs prior authorisation from a judicial authority, or an independent administrative authority whose decision is binding; in a duly justified situation of urgency, authorisation must be requested without undue delay and at the latest within 24 hours, and if it is rejected the use must stop immediately and all data, results, and outputs must be discarded and deleted (Article 5(3)). No decision that produces an adverse legal effect on a person may be taken based solely on the system’s output (Article 5(3)). The route exists only where a member state’s national law provides for it, and member states may adopt more restrictive laws (Article 5(5)).

## What did the Digital Omnibus add to Article 5?

The Article 5 prohibitions have applied since February 2, 2025, except the two practices the Omnibus added in Article 5(1)(ba) and (bb), which apply from December 2, 2026 (amended Article 113(a)).

New Article 5(1a) sets how far they reach. Placing such a system on the market or putting it into service is prohibited only where that generation or manipulation is the system’s intended purpose, or where the system’s design, training, architecture, capabilities, or user-facing functionalities make it a reasonably foreseeable and reproducible outcome, without requiring significant technical modification, and the system does not have reasonable and adequate technical safety measures and other safeguards to reliably prevent that generation or manipulation, taking into account reasonably foreseeable misuse, and to correct observed or reported misuse. Use is prohibited only where the deployer uses the system for the purpose of generating or manipulating such material or performance.

The second limb can reach a general-purpose image, video, or voice generator never marketed for this use. Recital 12 of the Omnibus lists safeguards that could qualify, including refusal training, output controls, content filtering, and abuse detection. Recital 13 says the exception covers activities where a “without right” defence applies under national law, such as the legitimate generation or manipulation of such material by the authorities to conduct criminal proceedings or to prevent, detect, or investigate crime, and the legitimate use of an AI system in red-teaming and evaluation to assess its compliance with the prohibition. Recital 15 adds that the same conduct may also violate other law, including criminal law, and that the prohibitions do not preclude prosecution under it.

## What do the Commission’s Article 5 guidelines say?

Article 96(1)(b) requires the Commission to issue guidelines on the prohibited practices. It approved the content of its “Guidelines on prohibited artificial intelligence practices established by Regulation (EU) 2024/1689 (AI Act)” and published them on February 4, 2025, stating that they would apply only once formally adopted (C(2025) 884 final), and formally adopted them on July 29, 2025 as Communication C(2025) 5052 final. The guidelines say they are non-binding and that any authoritative interpretation of the AI Act may ultimately only be given by the Court of Justice of the European Union (paragraph 5). The Commission’s page says they address practices such as harmful manipulation, social scoring, and real-time remote biometric identification.

The guidelines predate the Omnibus. As of November 2, 2026, the Commission’s page listed no version updated for the Omnibus, so for points (ba) and (bb) the text of Article 5(1a) and (1b) and the Omnibus recitals are the main guide.

## What do the Article 50 transparency obligations require, and what can a breach cost?

Article 50 applies from August 2, 2026, and a breach falls in the €15 million or 3% tier (Article 99(4)(g)).

- **Article 50(1), providers.** Providers must ensure that AI systems intended to interact directly with natural persons are designed and developed so that the people concerned are informed that they are interacting with an AI system, unless this is obvious from the point of view of a natural person who is reasonably well-informed, observant, and circumspect, taking into account the circumstances and the context of use.
- **Article 50(2), providers.** Providers of AI systems, including general-purpose AI systems, generating synthetic audio, image, video, or text content must ensure that the outputs are marked in a machine-readable format and detectable as artificially generated or manipulated. Providers must ensure their technical solutions are effective, interoperable, robust, and reliable as far as this is technically feasible, taking into account the specificities and limitations of various types of content, the costs of implementation, and the generally acknowledged state of the art. The duty does not apply to the extent the system performs an assistive function for standard editing or does not substantially alter the input data provided by the deployer or its semantics. Generative systems placed on the market before August 2, 2026 have until December 2, 2026 to meet the marking duty (Article 111(4)).
- **Article 50(3), deployers.** Deployers of an emotion recognition system or a biometric categorisation system must inform the natural persons exposed to it of its operation, and process the personal data in accordance with Regulations (EU) 2016/679 and (EU) 2018/1725 and Directive (EU) 2016/680, as applicable.
- **Article 50(4), deployers.** Deployers of a system that generates or manipulates image, audio, or video content constituting a deep fake must disclose that the content has been artificially generated or manipulated, except that, where the content forms part of an evidently artistic, creative, satirical, fictional, or analogous work or programme, the duty is limited to disclosing the existence of the generated or manipulated content in an appropriate manner that does not hamper the display or enjoyment of the work. Deployers of a system that generates or manipulates text published to inform the public on matters of public interest must disclose that the text has been artificially generated or manipulated, unless it has undergone human review or editorial control and a natural or legal person holds editorial responsibility for the publication.
The information must be given in a clear and distinguishable manner at the latest at the time of the first interaction or exposure, and must conform to the applicable accessibility requirements (Article 50(5)). Each paragraph also carries a law enforcement exception on its own terms.

## What happens if a provider misclassifies a high-risk system?

Where a market surveillance authority has sufficient reason to consider that a system classified as non-high-risk is high-risk, it must evaluate the classification against the Article 6(3) conditions and the Commission guidelines, and, if it finds the system high-risk, it must require the provider to bring the system into compliance and take appropriate corrective action within a period it may prescribe (Article 80(1) and (2)). A provider that fails to do so within the period set, or that misclassified the system to circumvent the Chapter III, Section 2 requirements, is subject to fines under Article 99 (Article 80(4) and (7)). Article 6(3) itself applies from December 2, 2027.

## Who fines EU institutions and general-purpose AI model providers?

The European Data Protection Supervisor fines Union institutions, bodies, offices, and agencies, up to €1.5 million for breaching Article 5 and up to €750,000 for other non-compliance (Article 100).

The Commission fines providers of general-purpose AI models up to 3% of annual total worldwide turnover in the preceding financial year or €15 million, whichever is higher, when it finds that the provider intentionally or negligently infringed the relevant provisions, failed to comply with a request for a document or information under Article 91 or supplied incorrect, incomplete, or misleading information, failed to comply with a measure requested under Article 93, or failed to give access to the model for an evaluation under Article 92 (Article 101(1)).

The model duties have applied since August 2, 2025. The AI Office’s powers to enforce them, including fines under Article 101, apply from August 2, 2026. A request for information is not a fine, but a provider that fails to comply with one, or supplies incorrect, incomplete, or misleading information, can be fined under Article 101.

The Omnibus also gave the AI Office fining powers over AI systems within its exclusive competence under Article 75(1): a non-compliance decision may carry penalties under Article 99(3) to (7), applied mutatis mutandis. For those systems the €15 million or 3% tier covers infringement of any applicable provision of the Act, including those not listed in Article 99(4), failure to comply with decisions or measures the AI Office adopts under Article 75a or Regulation (EU) 2019/1020, and failure to comply with commitments made binding under Article 75b, and supplying incorrect, incomplete, or misleading information to the AI Office in reply to a request falls in the €7.5 million or 1% tier (Article 75c(4)). The AI Office may also impose periodic penalty payments of up to 5% of average daily income or worldwide annual turnover in the preceding financial year per day (Article 75c(5)). These powers have no separate start date in Article 113, so they apply from the general date of August 2, 2026; the underlying duties keep their own dates.

## Who enforces the EU AI Act?

National market surveillance authorities enforce most of the Act. Each member state must designate at least one, plus a notifying authority and a single point of contact (Article 70(1) and (2)). For high-risk systems used by financial institutions in direct connection with financial services, the financial supervisor takes that role (Article 74(6)), and the European Data Protection Supervisor acts for EU institutions (Article 74(9)).

The AI Office is a Commission function (Article 3(47)). The Commission has exclusive powers to supervise and enforce the general-purpose AI model rules and entrusts them to the AI Office (Article 88(1)). As amended by the Omnibus, the AI Office is also exclusively competent for AI systems based on a general-purpose AI model where the model and the system come from the same provider or undertaking, subject to exceptions such as Annex I products and Annex III point 2 systems, and for AI systems that constitute or are integrated into a very large online platform or search engine (Article 75(1)).

Anyone with grounds to consider that the Act has been infringed may complain to the relevant market surveillance authority (Article 85).

## Have member states adopted their penalty laws?

Article 99(1) leaves the penalty rules to member states, and Article 99(2) required each to notify them to the Commission “without delay and at the latest by the date of entry into application.” Italy has adopted one. This guide has not surveyed the rest.

Italy’s Law No. 132 of September 23, 2025 designated the National Cybersecurity Agency (ACN) as market surveillance authority and single point of contact and the Agency for Digital Italy (AgID) as notifying authority, while the Bank of Italy, CONSOB, and IVASS keep the market surveillance role for financial institutions under Article 74(6) of the Act (Article 20(2)). Legislative Decree No. 179 of October 7, 2026, in force since October 23, 2026, then set the fines. Unless the conduct is a criminal offence, it applies the Article 99 ceilings, listing Articles 73 and 80 alongside the Article 99(4) duties (Article 23(1) to (3)). It adds a tier of up to €1 million or 0.5% of worldwide annual turnover, whichever is higher, for breaches of Article 27, the fundamental rights impact assessment, and Article 86, the right to explanation, which Article 99(4) does not list (Article 23(4)). For violations of minor harm or danger, judged in line with the Article 99(7) criteria and whether the obligation breached is substantive or formal, it lets the authorities, instead of fining, order the infringer to end the infringement and not repeat it or, where the infringement has ceased, impose a public declaration naming the violation and the party responsible; failing to obey such an order by the deadline brings a fine increased by up to one third, within the Article 23 ceilings (Articles 22 and 23(6)).

Other national laws can differ, so identify the authority and the penalty law in each member state where you operate.

## What should a company check first?

- Test every system placed on the EU market, or whose output is used in the EU, against Article 5(1), starting with emotion recognition at work or in education
- For any system that generates images, video, or audio of people, document before December 2, 2026 the safeguards that prevent non-consensual intimate material and child sexual abuse material, and how reported misuse is corrected
- Record whether the company is an SME or an SMC: SMEs get the lower figure in every Article 99 tier, SMCs only in the €15 million and €7.5 million tiers
How this fits the other guides: the [EU AI Act compliance guide](https://superintelligencecompliance.ai/guides/eu-ai-act-compliance) covers scope, risk tiers, and the Omnibus dates, the [EU AI Act compliance checklist](https://superintelligencecompliance.ai/guides/eu-ai-act-compliance-checklist) lists each duty by role with its article number, and the [high-risk classification guide](https://superintelligencecompliance.ai/guides/eu-ai-act-high-risk-ai-systems) covers Article 6, Annex III, and the Article 6(3) exception.

This guide is general information as of November 2026, not legal advice. Confirm each point against the consolidated text of Regulation (EU) 2024/1689, the national penalty law of each member state concerned, and with EU counsel.

## Sources

- [Regulation (EU) 2024/1689 (Artificial Intelligence Act)](https://eur-lex.europa.eu/eli/reg/2024/1689/oj), Official Journal of the European Union, July 12, 2024.
- [Regulation (EU) 2024/1689, consolidated text as of July 27, 2026](https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:02024R1689-20260727), EUR-Lex, Publications Office of the European Union, accessed November 2, 2026.
- [Regulation (EU) 2026/1744 (Digital Omnibus on AI)](https://eur-lex.europa.eu/eli/reg/2026/1744/oj?locale=en), Official Journal of the European Union, July 24, 2026.
- [Directive 2011/93/EU on combating the sexual abuse and sexual exploitation of children and child pornography](https://eur-lex.europa.eu/eli/dir/2011/93/oj), Official Journal of the European Union, December 17, 2011.
- [Commission publishes the Guidelines on prohibited artificial intelligence (AI) practices, as defined by the AI Act](https://digital-strategy.ec.europa.eu/en/library/commission-publishes-guidelines-prohibited-artificial-intelligence-ai-practices-defined-ai-act), European Commission, February 4, 2025.
- [Communication from the Commission, Commission Guidelines on prohibited artificial intelligence practices established by Regulation (EU) 2024/1689 (AI Act), C(2025) 5052 final](https://ec.europa.eu/newsroom/dae/redirection/document/112367), European Commission, July 29, 2025.
- [Communication to the Commission, Approval of the content of the draft Communication from the Commission, Commission Guidelines on prohibited artificial intelligence practices established by Regulation (EU) 2024/1689 (AI Act), C(2025) 884 final](https://ec.europa.eu/newsroom/dae/redirection/document/112366), European Commission, February 4, 2025.
- [Legge 23 settembre 2025, n. 132, Disposizioni e deleghe al Governo in materia di intelligenza artificiale](https://www.gazzettaufficiale.it/eli/id/2025/09/25/25G00143/sg), Gazzetta Ufficiale della Repubblica Italiana, Serie Generale n. 223, September 25, 2025.
- [Decreto legislativo 7 ottobre 2026, n. 179, Adeguamento della normativa nazionale alle disposizioni del regolamento (UE) 2024/1689 del Parlamento europeo e del Consiglio, del 13 giugno 2024, che stabilisce regole armonizzate sull’intelligenza artificiale, in materia di poteri delle Autorità nazionali e di utilizzo dell’intelligenza artificiale nella formazione](https://www.gazzettaufficiale.it/eli/id/2026/10/08/26G00200/sg), Gazzetta Ufficiale della Repubblica Italiana, Serie Generale n. 234, October 8, 2026.

## Common questions

**What are the maximum fines under the EU AI Act?**

Under Article 99, up to €35 million or 7% of total worldwide annual turnover for the preceding financial year, whichever is higher, for non-compliance with the Article 5 prohibitions; up to €15 million or 3%, whichever is higher, for breaches of the operator and notified body obligations listed in Article 99(4), including the Article 50 transparency obligations; and up to €7.5 million or 1%, whichever is higher, for supplying incorrect, incomplete, or misleading information to notified bodies or national competent authorities in reply to a request. For SMEs, including start-ups, the lower figure applies, and for small mid-cap enterprises the lower figure applies to the second and third tiers only (Article 99(6) and (6a)).

**Which AI practices are prohibited under the EU AI Act?**

Article 5(1) bans subliminal, purposefully manipulative, or deceptive techniques, and the exploitation of vulnerabilities due to age, disability, or a specific social or economic situation, where they materially distort behaviour in a way that causes or is reasonably likely to cause significant harm; social scoring that leads to unjustified or unrelated detrimental treatment; predicting a person’s risk of committing a criminal offence based solely on profiling or on assessing personality traits (except to support a human assessment already based on objective and verifiable facts directly linked to a criminal activity); untargeted scraping of facial images from the internet or CCTV footage to build facial recognition databases; emotion recognition in the workplace and education except for medical or safety reasons; biometric categorisation to deduce or infer race, political opinions, trade union membership, religious or philosophical beliefs, sex life, or sexual orientation, except labelling or filtering lawfully acquired biometric datasets and categorising biometric data in the area of law enforcement; and real-time remote biometric identification in publicly accessible spaces for law enforcement except under strict conditions. These have applied since February 2, 2025. Two prohibitions added by Regulation (EU) 2026/1744, on non-consensual intimate material and child sexual abuse material, apply from December 2, 2026.

**Who imposes EU AI Act fines?**

Member states lay down the penalty rules (Article 99(1)), and national market surveillance authorities enforce most of the Act; in some legal systems a court imposes the fine (Article 99(9)). The Commission, through the AI Office, fines providers of general-purpose AI models under Article 101, a power that applies from August 2, 2026; since the Digital Omnibus, the AI Office can also fine providers of AI systems within its exclusive competence under Article 75(1) (Article 75c); and the European Data Protection Supervisor fines EU institutions, bodies, offices, and agencies under Article 100.

**When do the EU AI Act penalties apply?**

Chapter XII, which contains the penalty articles, has applied since August 2, 2025, except Article 101 on fines for general-purpose AI model providers, which applies from August 2, 2026 (Article 113(b)). The underlying duties have their own dates: the Article 50 transparency duties apply from August 2, 2026, the two Omnibus prohibitions from December 2, 2026, and the high-risk duties from December 2, 2027 for stand-alone Annex III systems and August 2, 2028 for AI in Annex I Section A regulated products.