The Chapter III high risk rules for AI systems classified under Article 6(1) and Annex I apply from 2 August 2028 under Article 113 as amended by Regulation (EU) 2026/1744. That is the product conformity timeline that matters most in this cluster. Earlier dates apply to other parts of the Act: the Regulation entered into force on 1 August 2024, prohibited practices and AI literacy provisions applied from 2 February 2025, general purpose AI model rules applied from 2 August 2025, and the majority of rules applied from 2 August 2026. Transitional provisions in Article 111 and in the sector legislation itself can change the position for systems already placed on the market, and must be checked separately for each product.
What is the difference between entry into force and application?
Entry into force and application are different events, and confusing them causes timeline errors. Regulation (EU) 2024/1689 entered into force on 1 August 2024. That means the Regulation became part of Union law on that date. Application dates in Article 113 then set when particular chapters and provisions start to apply in practice. A provision that has entered into force but has not yet reached its application date does not yet impose the duties it describes, although preparatory work, guidance and standards work can properly begin before the date. A provision whose application date has passed imposes its duties from that date, subject to transitional provisions. The whole Act did not come into force on 2 August 2026, and it did not start to apply as a single block on any one date. Every date claim in a product plan should therefore name the provision, the instrument that set the date and whether the date is an application date, a transitional deadline or a sector date. A plan that lists dates without provisions will be difficult to audit and easy to misread.
Which earlier dates affect product teams?
Product teams are affected by earlier dates even though the Annex I high risk rules apply from 2 August 2028. From 2 February 2025, the general provisions including AI literacy in Article 4 and the original prohibited practices in Article 5 applied. From 2 August 2025, the general purpose AI model rules in Chapter V applied, together with governance arrangements. From 2 August 2026, the majority of rules applied, including Article 50 transparency rules, and enforcement started for the rules then applicable. From 2 December 2026, the two new prohibitions in Article 5(1)(ba) and (bb) added by Regulation (EU) 2026/1744 apply, together with the transitional deadline for certain providers of synthetic content systems already placed on the market before 2 August 2026 to comply with Article 50(2). A product team that supplies a general purpose model, that provides a system that interacts with users, or that generates synthetic content must therefore meet duties well before 2 August 2028, even where the product high risk rules are later. The product plan should list the duties by provision and date, not by a single product launch date.
For the broader regulatory framework, see products, machinery and vehicles overview.
What does 2 August 2028 mean for Annex I products?
From 2 August 2028, Chapter III, Sections 1 to 3 apply to AI systems classified as high risk under Article 6(1) and Annex I, with the exception of Article 6(5) which is treated separately in Article 113. In practical terms, that means the high risk requirements in Chapter III, Section 2, the provider and deployer obligations in Section 3, and the conformity, registration and post market provisions that attach to them, apply to Annex I products from that date. The date was 2 August 2027 in the original Article 113, and Regulation (EU) 2026/1744 moved it to 2 August 2028. The move gives more time for standards, sector integration and conformity capacity, but it does not remove the duties and it does not extend the sector duties that apply under the sector legislation itself. A product placed on the market before 2 August 2028 will need a transition analysis under Article 111 and under the sector legislation, and a product placed on the market after that date will need the full file from the outset. The plan should work backwards from 2 August 2028, allowing time for standards, notified body capacity, testing and documentation, rather than treating the date as a start date for preparation.
What transitional provisions must be checked?
Article 111 contains transitional provisions for high risk systems already placed on the market or put into service before the relevant application dates, and for general purpose AI models placed on the market before 2 August 2025. Those provisions are detailed and conditional, and they differ between Annex I and Annex III systems, between systems that undergo substantial modification and systems that do not, and between public authority systems and other systems. This guide does not reproduce Article 111 in full, and a product team should not rely on a summary of it for a transition decision. The sector legislation also contains its own transitional provisions, for example for machinery and for medical devices, which can affect the same product at the same time. The transition record for a product should cite the Article 111 provision relied on, the sector transition provision relied on, the date the product was placed on the market, any substantial modification since that date, and the conclusion with reasons. Where a product is substantially modified after the application date, the transition protection can be lost, and that consequence should be assessed before the modification is released. Transitional provisions are not a general grace period.
How should a product team plan backwards from 2028?
A backwards plan from 2 August 2028 should start with the conformity file, not with the product launch. It should identify the sector legislation, the conformity procedure, the notified body if required, the harmonised standards or common specifications to be applied, the test and validation evidence, the technical documentation, the declaration and the registration and post market arrangements. It should then identify the dependencies: standards that are not yet cited in the Official Journal, sector delegated acts that are not yet adopted, notified body capacity, and data and evaluation work that takes calendar time. Each dependency should have an owner, a check date and a fallback, such as demonstrating compliance by other means where the Act allows it, with the reasoning recorded. The plan should include a freeze point for design changes, a test window, a documentation review and a pre market review by a person who was not responsible for the design. It should also include the earlier dates described above, so that transparency, literacy and general purpose model duties are not discovered late. A plan that lists only the 2028 date will miss the work that must be done in 2026 and 2027.
Continue with Annex I Section A and Section B.
What does this timeline guide not settle?
This guide does not settle the transitional position for any named product, does not predict standards or sector delegated acts, and does not state the application dates for any sector legislation. It uses only the verified date list for the AI Act itself, and every other date must be checked in the instrument that sets it. A product team with a specific product and a specific placing on the market date should confirm the Article 111 position, the sector transition position and the current standards position with a competent authority or legal adviser. Timeline planning is not a substitute for a legal determination, and a date that has moved once, as the Annex I date did under Regulation (EU) 2026/1744, can move again only by a further legislative act, which should not be assumed.
Frequently asked questions
When do Annex I high risk rules apply?
From 2 August 2028, under Article 113 as amended by Regulation (EU) 2026/1744, for AI systems classified as high risk under Article 6(1) and Annex I. That date is distinct from the 2 December 2027 date for Annex III systems and from the earlier dates for prohibitions, general purpose models and transparency duties.
Did the whole AI Act come into force on 2 August 2026?
No. The Regulation entered into force on 1 August 2024. 2 August 2026 is the date from which the majority of rules applied, including Article 50 transparency rules. Different provisions have different application dates, and each date claim should name the provision it refers to.
What is the 2 December 2026 date?
On 2 December 2026, the two new prohibitions in Article 5(1)(ba) and (bb) added by Regulation (EU) 2026/1744 apply, and the transitional deadline ends for certain providers of synthetic content systems already placed on the market before 2 August 2026 to comply with the Article 50(2) marking and detection duty. Product teams with synthetic content functions should check that date separately from the 2028 product date.
Do transitional provisions give a general grace period?
No. Article 111 transitional provisions are detailed and conditional, and they differ by system type, placing date and modification history. Sector transitional provisions must also be checked. A transition decision needs the specific provisions cited, the placing date and the modification history, confirmed with a competent authority or legal adviser.
Continue through this AI Act cluster
This article is an informational overview, not a legal determination. Check the current legal text and official implementation guidance for your system and role.
Consolidated Regulation (EU) 2024/1689 ↗AI Act Service Desk timeline ↗