EU AI Act key dates: the 2024-2028 timeline
The main application dates of Regulation (EU) 2024/1689 as published in the Official Journal of the EU, with the changes made by Regulation (EU) 2026/1744. For each date: what applies, who it affects and what to prepare.
Content reviewed on
In short
The AI Act has been in force since 1 August 2024 and applies in stages. Prohibited practices and AI literacy from 2 February 2025; general-purpose AI models (with until 2027 for existing ones), governance and Article 99 penalties from 2 August 2025; general application, including Article 50 transparency, from 2 August 2026. Two new prohibitions follow on 2 December 2026. After Regulation (EU) 2026/1744, high-risk obligations apply from 2 December 2027 (Annex III) and 2 August 2028 (Annex I).
Next date
New prohibitions and end of the marking grace period
37 days to go
Important: This guide is educational and indicative information. It is not individual legal advice or a conformity assessment. For a specific case, consult a specialised law firm.
Application timeline, date by date
Official Journal dates. The Regulation has been in force since 2024; each date changes what applies. «Applies» means the date has arrived; «Pending» means not yet.
Applies
Entry into force
Regulation (EU) 2024/1689 enters into force twenty days after its publication in the Official Journal (12 July 2024). The transitional period begins: most obligations do not apply yet.
- Who it affects
- Providers, deployers (anyone using an AI system in their activity), importers and distributors within the scope of Article 2, with its exclusions.
- What to have in place
- Entry into force did not bring forward obligations with later dates. As preparation, list the AI tools and uses in your company.
Applies
Prohibited practices and AI literacy
Chapters I and II apply: the Article 5 prohibitions and Article 4 AI literacy. Since the July 2026 reform, Article 4 asks for tailored measures that support the AI literacy of staff and of anyone using the systems on the organisation's behalf, without setting a specific individual level.
- Who it affects
- Providers and deployers, SMEs included. The prohibitions concern the practices described in Article 5.
- What to have in place
- Check that no use falls under Article 5 and adopt tailored AI literacy measures for those who use the systems: training, instructions or a use policy can serve. No specific individual level is required.
Applies
General-purpose AI models, governance and fines
Chapter V obligations apply to providers of general-purpose AI models (documentation, information for those who integrate them, copyright policy), with additional obligations for models with systemic risk. Models already on the market have until 2 August 2027 to comply. Governance (AI Office, national authorities) and the Article 99 penalty regime also apply; Article 101 fines on general-purpose model providers apply from August 2026.
- Who it affects
- Providers of general-purpose AI models and their authorised representatives, and the authorities. From this date, failing to meet obligations that already apply can be penalised.
- What to have in place
- Using a tool does not make your company the model's provider. Be clear about your company's role and keep complying with what already applies.
Applies
Regulation (EU) 2026/1744 (Digital Omnibus on AI) enters into force
Published in the Official Journal on 24 July 2026, it postpones the high-risk obligations, adds two prohibited practices, grants a transitional period for marking AI-generated content and applies Articles 102 to 110 (adaptation of other EU laws) from this day.
- Who it affects
- All operators: the Regulation's timeline changes.
- What to have in place
- Review the new high-risk deadlines, the new wording of Article 4 and the two prohibitions that apply from 2 December 2026.
Applies
General application of the Regulation
General date of application, including Article 50 and Article 101 fines. Transparency splits the duties: providers must disclose that people are talking to an AI and mark synthetic content; deployers must, among other cases, disclose deepfakes. Generative systems already on the market have a special deadline for marking. High-risk obligations have later dates.
- Who it affects
- Providers of systems that interact with people or generate content and, in the cases of Article 50.3 and 50.4, deployers (for example, whoever uses an AI system to generate deepfakes they publish).
- What to have in place
- Work out which part of Article 50 applies and to whom. If you publish deepfakes, disclose it clearly. If you use a third-party chatbot, check with the provider how it tells people they are talking to an AI.
Pending
New prohibitions and end of the marking grace period
Two prohibited practices added to Article 5 apply, with their conditions: systems that generate or manipulate realistic intimate or sexually explicit depictions of an identifiable person without consent, and child sexual abuse material. The special deadline for Article 50.2 marking also ends for generative systems placed on the market before 2 August 2026.
- Who it affects
- Anyone carrying out the prohibited practices under the conditions of the amended Article 5, and providers of generative systems placed on the market before 2 August 2026 (Article 50.2 marking).
- What to have in place
- Marking is the system provider's obligation. If your company only uses the tool, ask the provider how it complies and review your own transparency duties separately.
Pending
Annex III high risk
High-risk obligations apply to Annex III systems: certain uses in employment, education, creditworthiness of natural persons, essential services and other areas, with exceptions. Providers meet requirements such as risk management and technical documentation; deployers have the Article 26 obligations (human oversight, monitoring, logs).
- Who it affects
- Providers, deployers and other operators of systems classified as high-risk under Article 6 and Annex III: certain uses in recruitment, education, creditworthiness of natural persons and essential services, with exceptions.
- What to have in place
- Identify uses that may be high-risk and check their classification. If Article 26 applies, prepare human oversight, monitoring, retention of the automatic logs under your company's control and information to the people affected. Systems placed on the market earlier have a transitional regime (Article 111.2).
Pending
Annex I high risk
High-risk obligations apply to Article 6.1 systems linked to regulated Annex I products such as toys, lifts or medical devices. After the reform, machinery moves to its own sectoral regime (Annex I, Section B).
- Who it affects
- Operators of systems meeting the Article 6.1 high-risk conditions, such as certain products or safety components of toys, lifts and medical devices.
- What to have in place
- Check which regime applies to the product and your company's role. If the system falls under the general high-risk regime (Annex I, Section A) and your company deploys it, following the manufacturer's instructions is only part of Article 26; Section B products follow their own sectoral law.
What the Digital Omnibus (Regulation 2026/1744) changed
Published in the Official Journal on 24 July 2026 and in force since 27 July. These are the main date changes.
| Obligation | Original date | Current date |
|---|---|---|
| Annex III high-risk (Chapter III, Sections 1 to 3, except Article 6.5)Original date: 2 August 2026. Current date: 2 December 2027 | 2 August 2026 | 2 December 2027 |
| Annex I high-risk, Section A (Article 6.1)Original date: 2 August 2027. Current date: 2 August 2028 | 2 August 2027 | 2 August 2028 |
| Article 50.2 marking for systems placed on the market before 2 August 2026Original date: 2 August 2026. Current date: 2 December 2026 | 2 August 2026 | 2 December 2026 |
| Two new Article 5 prohibitions (non-consensual intimate content and child sexual abuse)Original date: Not in Article 5. Current date: 2 December 2026 | Not in Article 5 | 2 December 2026 |
| Articles 102 to 110 (amendments to other EU laws)Original date: 2 August 2026. Current date: 27 July 2026 | 2 August 2026 | 27 July 2026 |
Frequently asked questions about the dates
When does the AI Act apply to SMEs?
It depends on your company's role and how it uses AI. Prohibited practices and AI literacy apply from 2 February 2025; Article 50 transparency from 2 August 2026 in its specific cases. High-risk obligations can come from Annex III (from 2 December 2027) or Annex I (from 2 August 2028), and each system's classification has to be checked.
Has the AI Act been postponed?
Only in part. Regulation (EU) 2026/1744 postponed the high-risk obligations (Annex III to 2 December 2027 and Annex I to 2 August 2028) and gave until 2 December 2026 for Article 50.2 marking, only for systems already on the market. The rest of Article 50 keeps its date, the prohibitions were not postponed, Article 4 was reworded and two new prohibitions were added.
Since when are there fines?
The Article 99 penalty regime applies from 2 August 2025, and Article 101, for general-purpose model providers, from 2 August 2026. Only obligations that already apply can be penalised. For SMEs, Article 99.6 uses the lower of the two maximum limits; the actual fine depends on the case.
What should an SME using ChatGPT or Copilot have in place today?
The tool's brand does not decide it. Know what it is used for and your company's role, check that no use is prohibited, have AI literacy measures for the people using it and see which part of Article 50 concerns you. This is guidance, not a full assessment.
How do I add these dates to my calendar?
The «Add the dates to my calendar» button downloads an .ics file with the dates on this page, which you can import into Google Calendar, Outlook or your phone's calendar.
Sources
The texts we read to write this guide. When in doubt, the regulation prevails.
Which dates affect you?
The free check gives you a first indication of which obligations could affect you based on how you use AI. It does not replace an analysis of your case.