AI Act 2 August 2026: What Applies from the General Application Date and What Moved to 2027 and 2028

Since 2 August 2026, the general date of application in Article 113, second paragraph, the EU AI Act applies Article 50 transparency, the Commission's fining powers over general-purpose AI model providers and the market surveillance and remedies chapters, while the high-risk rules wait for 2027 and 2028. The date covers seven chapters, Chapter XIII only in part, and Article 101, and the 2026 amendment did not move it.

The act is Regulation (EU) 2024/1689 (OJ L, 2024/1689, 12.7.2024), in force since 1 August 2024. Under the EU AI Act, the applicable text is Regulation (EU) 2024/1689 as amended by Regulation (EU) 2026/1744, in force since 27 July 2026, six days before the general date.

Chapter III, Sections 1 to 3, the core of the high-risk rules, did not apply on that date. Enforcement started that day: the Commission announced on 31 July 2026 (IP/26/1714) that the AI Office and national authorities enforce the applicable rules from 2 August 2026.

Compiled by the AIActGuide.net Editorial Team from the Official Journal text; how we check every statement: methodology.

What Applies under the AI Act since 2 August 2026?

Article 113, second paragraph, applies every provision not covered by points (a) to (d) of its third paragraph from 2 August 2026: Chapter IV (Article 50), Chapters VI, VIII, IX, X, XI and XIII and Article 101. The provisions that became applicable on that date are these:

The earlier dates of the EU AI Act timeline cover the rest: Chapters I and II (definitions, AI literacy, prohibitions) since 2 February 2025 under point (a), and Chapter III, Section 4, Chapters V, VII and XII and Article 78 since 2 August 2025 under point (b), Article 101 excepted. AI Act 2 August 2025 was the milestone for GPAI obligations, governance and penalties.

Within Chapter XIII, Articles 102 to 110, which amend sectoral Union acts, did not wait for the general date. Point (d), added by Regulation (EU) 2026/1744, applies them from 27 July 2026, the day the amending act entered into force.

What Providers and Deployers Must Disclose since 2 August 2026

Article 50 has applied since 2 August 2026: providers must disclose AI interaction (Article 50(1)) and mark synthetic output in a machine-readable, detectable format (Article 50(2)), and deployers must disclose emotion recognition, biometric categorisation and deep fakes (Article 50(3) and (4)). Article 50(2) covers systems generating synthetic audio, image, video or text, including general-purpose AI systems.

Deployer disclosure reaches further than deep fakes. Under the transparency obligations under Article 50 for deployers, a deep fake is image, audio or video content only (Article 3(60)), and Article 50(4) also covers AI-generated text published to inform the public on matters of public interest. Article 50(5) requires clear and distinguishable information at the latest at the first interaction or exposure, conforming to accessibility requirements.

The duties carry exceptions, for example where the interaction is obvious to a reasonably well-informed, observant and circumspect natural person (Article 50(1)) or where a system performs an assistive function for standard editing (Article 50(2)).

Only Article 50(2) has a transition: generative systems placed on the market before 2 August 2026 comply by 2 December 2026 (Article 111(4)).

The Code of Practice on Transparency of AI-generated Content, final since 10 June 2026 and found adequate by the Commission on 8 July 2026, covers Article 50(2), (4) and (5), including the marking and detection of AI-generated content and deployer labelling. Adherence to the code is not conclusive evidence of compliance, and IP/26/1714 of 31 July 2026 counted "more than 180 organisations" among its first signatories.

The Commission guidelines on Article 50 (content approved and published on 20 July 2026) confirm in paragraph 153 that Article 50(1) applies from 2 August 2026 also to partly interactive systems. Footnote 50 of the guidelines states that the Article 111(2) grandfathering does not reach Article 50.

The Commission's Supervisory Powers over Model Providers from 2 August 2026

The GPAI model obligations of Chapter V have applied since 2 August 2025, but the Commission's powers to supervise and enforce them (Chapter IX, Section 5, Articles 88 to 94) and to fine providers under Article 101 apply only since 2 August 2026. Article 88(1) gives the Commission exclusive powers over Chapter V and entrusts them to the AI Office.

For the obligations of general-purpose AI models under the AI Act, the Commission can request documents (Article 91), evaluate a model (Article 92) and require compliance, systemic-risk mitigation, restriction, withdrawal or recall (Article 93). Article 101 fines apply where the provider intentionally or negligently infringed the relevant provisions, failed to comply with an Article 91 request or supplied incorrect, incomplete or misleading information, failed to comply with an Article 93 measure, or failed to give access for an Article 92 evaluation, and Implementing Regulation (EU) 2026/1755 (OJ L, 2026/1755, 21.7.2026) sets the procedure.

Executive Vice-President Virkkunen said on 14 October 2026 (SPEECH/26/2150) that the Commission "has already taken our first enforcement actions", with requests for information on security, external evaluation and monitoring how models are used. Models placed on the market before 2 August 2025 have until 2 August 2027 to comply (Article 111(3)).

Chapter IX Procedures and the Other Provisions That Became Applicable

Chapter IX, which holds the market surveillance, enforcement, remedies and GPAI supervision rules (Articles 72 to 94, with Articles 75a to 75d), has applied since 2 August 2026. Article 75a, inserted by Regulation (EU) 2026/1744, gives the AI Office the powers of a market surveillance authority for the systems under its competence, including requests for information and remote or on-site inspections. It acts on its own initiative, on an Article 85 complaint or at the request of a market surveillance authority.

Article 85 lets any natural or legal person complain to a market surveillance authority, without having to be personally affected; in our reading, Article 85 gives no right to a particular outcome. The Commission runs a Complaints Tool, a Whistleblower Tool and a channel for downstream providers, and these sit beside the other rights of affected persons under Articles 85 to 87.

Chapter VI also lets the AI Office establish a Union-level AI regulatory sandbox for AI systems under its exclusive competence (Article 57(3a)).

Which Authority Enforces the Rules for Each Kind of System since 2 August 2026

Enforcement is split three ways since 2 August 2026: the AI Office for GPAI models, for AI systems built on a GPAI model by the same provider and for systems in designated very large online platforms and search engines; national competent authorities for all other AI systems; the European Data Protection Supervisor for Union institutions (IP/26/1714, 31 July 2026). The competences rest on these provisions:

The AI Office competence reaches deployers only when they are also the provider or part of the same undertaking (Article 75(1), second subparagraph). Everything else falls to the AI Act national competent authorities: Article 70 required Member States to designate a notifying authority, a market surveillance authority and a single point of contact by 2 August 2025.

IP/26/1714 also names a scientific panel of 60 independent AI experts, with Prof. Alessandro Abate (University of Oxford) as lead scientific adviser.

What Did Not Apply on 2 August 2026, and Where It Moved

The high-risk rules of Chapter III, Sections 1 to 3, did not apply on 2 August 2026: they apply from 2 December 2027 to Annex III systems and from 2 August 2028 to Annex I systems (Article 113, third paragraph, point (c)(i) and (ii), as amended). Regulation (EU) 2026/1744 changed these dates:

The moves differ in length. For the classification and requirements of high-risk AI systems, Annex III systems gained 16 months after 2 August 2026 and Annex I systems 12 months after 2 August 2027. Article 6(5), the Commission's duty to issue classification guidelines, is excluded from the postponement.

Sections 1 to 3 are postponed. The national sandbox deadline in Article 57(1) and the Commission's post-market monitoring template under Article 72(3) moved separately, by their own amended provisions.

How Regulation (EU) 2026/1744 Changed the Calendar Six Days Earlier

Regulation (EU) 2026/1744 entered into force on 27 July 2026, six days before the general date of application, after Parliament's position of 16 June 2026 and Council adoption on 29 June 2026. The Commission proposed it as COM(2025) 836 on 19 November 2025. Parliament adopted its position by 423 votes to 57 with 174 abstentions, and the act was signed on 8 July 2026 and published in OJ L, 2026/1744, 24.7.2026.

Recital 40 explains the move. The Digital Omnibus on AI gives its reasons there: delayed harmonised standards, common specifications, guidance and national authorities.

The proposal's "stop-the-clock" mechanism (a Commission decision followed by 6 or 12 months) was not adopted, and the final text sets fixed dates. The general date of 2 August 2026 and the Article 50 and GPAI dates stayed unchanged; only the Article 111(4) transition was new, shortened from the proposed 2 February 2027 to four months (recital 38).

Maximum Amounts for Breaches of the Rules That Apply since 2 August 2026

Breaches of Article 50 carry fines of up to EUR 15 000 000 or 3 % of total worldwide annual turnover, whichever is higher (Article 99(4), point (g)), and the Commission can fine GPAI model providers up to 3 % or EUR 15 000 000, whichever is higher (Article 101(1)). The four tiers sit side by side:

For SMEs, including start-ups, each Article 99 maximum is the lower of the two amounts (Article 99(6)); for small mid-caps the lower amount applies under Article 99(4) and (5) (Article 99(6a)). For AI systems under its exclusive competence, the AI Office itself imposes fines, applying Article 99(3) to (7) mutatis mutandis (Article 75c(4)).

The Article 99 amounts are ceilings. Member States lay down the rules on EU AI Act penalties under Article 99(1), in Chapter XII since 2 August 2025, and Article 99 sets the maximum fines those national rules must provide for.

Commitments under an adequate code of practice may count as a mitigating factor (Article 99(7), point (e); paragraph 149 of the Article 50 guidelines). Article 4 on AI literacy is not listed in Article 99(4), so the act sets no EU fine tier for it.

What Providers and Deployers Should Have in Place since 2 August 2026

Operators should by now have mapped every AI system to the Article 50 duties, the GPAI rules and the high-risk dates that apply to it. Six steps give an orientation, not an assessment of any individual system:

1. List every AI system and GPAI model, the role held for each and its placing-on-the-market date.
2. Disclose AI interaction at the latest at the first interaction (Article 50(1) and (5)).
3. Implement machine-readable marking and detection for generative systems, with legacy systems meeting the Article 111(4) deadline (Article 50(2)).
4. Label deep fakes and AI-generated text published on matters of public interest as a deployer (Article 50(4)).
5. Keep GPAI technical documentation ready for Commission requests (Articles 91 and 101).
6. Plan the high-risk work for the Annex III or Annex I date with the Commission's draft classification guidelines of 19 May 2026; the Commission announced on 19 May 2026 that the final guidelines will be adopted by the end of 2026.

The steps follow the act's roles. The EU AI Act checklist sorts obligations by provider, deployer and the other operator roles, and by date. Step 5 concerns only providers of general-purpose AI models.

A generative system placed on the market since 2 August 2026 is bound by Article 50(2) from its first day, because Article 111(4) covers only earlier systems.

After 2 August 2026: The Next AI Act Milestones

The next fixed date is 2 December 2026, when the new prohibitions in Article 5(1), points (ba) and (bb), and the Article 50(2) deadline for generative systems placed before 2 August 2026 apply. Each later date rests on Article 113 or a transitional provision, as the overview of every application date of the EU AI Act shows.

AI Act 2 August 2027 follows, with legacy GPAI models (Article 111(3)), the national sandbox deadline and the Article 2(13) delegated acts, before the high-risk milestones AI Act 2 December 2027 and AI Act 2 August 2028. Between them, Annex I Section A notified bodies must apply for designation by 28 January 2028 (Article 43(3)).

Points (a), (b) and (c) to (h) of Article 5(1) apply since 2 February 2025, points (ba) and (bb) with Article 5(1a) and (1b) from 2 December 2026. The page on AI Act 2 December 2026 covers both new prohibitions.

AI Act 2 August 2026: Related Questions

Article 2(1), Article 4 and Article 113 of the AI Act, with the Commission's Article 50 guidelines and enforcement page, settle the common questions about 2 August 2026.

Did the Whole AI Act Apply from 2 August 2026?

No: 2 August 2026 is the general date, but Chapters I and II and parts of Chapter XIII applied earlier, and Chapter III, Sections 1 to 3, apply only from the later high-risk dates.

No single date switched on the whole text. The EU AI Act summary explains chapter by chapter what each part regulates.

Does the 2 August 2026 Milestone Reach Companies outside the EU?

Yes: the rules applicable since 2 August 2026 bind providers established outside the EU when they place AI systems on the EU market or when the output is used in the EU (Article 2(1), points (a) and (c)).

Third-country deployers are covered on the same output test. The guide to the EU AI Act for non-EU companies sets out when an authorised representative is needed.

Must Content Generated before 2 August 2026 Be Labelled?

No retroactive labelling is required for content generated before 2 August 2026, but pre-existing texts published on or after that date must be labelled, according to the Commission's Article 50 guidelines (content approved and published on 20 July 2026, paragraph 154).

The publication date decides. The labelling duties of deployers under Article 50(4) cover deep fakes and public-interest text.

Is AI Literacy Supervised since 2 August 2026?

Yes: the Commission's AI Literacy Q&A (last update 20 October 2026) states that national market surveillance authorities, not the AI Office, supervise Article 4 from 2 August 2026.

Since 27 July 2026, Article 4 requires providers and deployers to "take measures to support the development of AI literacy". The duty of AI literacy under Article 4 does not require a guaranteed level of AI literacy for any individual.

Has the Commission Published a Fining Decision since 2 August 2026?

No AI Act fining decision is published by the Commission on its enforcement page as of 24 October 2026, while Executive Vice-President Virkkunen confirmed on 14 October 2026 that the AI Office has sent requests for information.

Three routes lead to a fine. EU AI Act penalties and fines are imposed under national rules (Article 99), by the Commission on GPAI model providers (Article 101) or by the AI Office for systems under its exclusive competence (Article 75c(4)).

Where Is the Text of the AI Act as It Applies in 2026?

The consolidated text of 27 July 2026 (02024R1689-20260727) shows the AI Act with the amendments of Regulation (EU) 2026/1744, but only the two Official Journal publications are authentic.

The consolidation has 119 articles and 14 annexes, and the EU AI Act full text points to both publications: OJ L, 2024/1689, 12.7.2024 and OJ L, 2026/1744, 24.7.2026.