The EU AI Act deadlines
The EU AI Act does not apply all at once, but in clearly defined stages. Since 2 August 2026 the Regulation has applied generally. The Digital Omnibus (Regulation (EU) 2026/1744) has postponed the deadlines for high-risk AI to 2 December 2027 and 2 August 2028.
The deadline passed long ago. You can get started today.
AI literacy has been mandatory since February 2025. Our training takes the work off your hands: invite your staff in one step, each person receives a certificate, your company the audit report as CSV or PDF.
from €24.69 per person, net of VAT (no subscription)
The EU AI Act timeline
The EU AI Act entered into force on 1 August 2024; its provisions apply in stages from 2 February 2025 to 2 August 2028. The prohibitions apply earlier because they concern unacceptable risks (recital 179); for high-risk AI, the Omnibus has postponed the dates to allow time for adaptation and implementation (recital 40 of Regulation (EU) 2026/1744). Those who wait until all deadlines have passed, however, will have waited too long. The timeline below shows which obligations apply from when:
- 1 August 2024: The EU AI Act (Regulation (EU) 2024/1689) enters into force and is directly applicable in all member states as an EU Regulation, without national implementing legislation. The transitional periods for most obligations begin to run.
- 2 February 2025: Prohibited AI practices under Article 5 are banned with immediate effect. At the same time, the AI literacy obligation under Article 4 applies, initially in its original version: providers and deployers take measures to ensure, to their best extent, a sufficient level of AI literacy among their staff. Since 27 July 2026, Article 4 has applied in the version introduced by the Digital Omnibus.
- 2 August 2025: Obligations for GPAI models (General Purpose AI) take effect; for models placed on the market before this date, providers have until 2 August 2027 (Art. 111(3)). In parallel, the governance structure becomes applicable: the EU AI Office, national market surveillance authorities and the penalty provisions become enforceable, except for the fines on providers of GPAI models (Art. 101).
- 27 July 2026: Regulation (EU) 2026/1744 (Digital Omnibus on AI) enters into force. It rewords Article 4 and postpones the high-risk obligations. They were originally due on 2 August 2026 (Annex III) and 2 August 2027 (Annex I).
- 2 August 2026: The Regulation applies generally (Art. 113). This includes the transparency obligations under Article 50, for example for chatbots and AI-generated content, and the Commission's fines on providers of GPAI models under Art. 101.
- 2 December 2026: For generative AI systems placed on the market before 2 August 2026, the labelling obligation under Article 50(2) applies only from this date (transitional period). In addition, two new prohibitions under Article 5 apply: AI systems that generate non-consensual intimate depictions of real persons, and AI systems that generate child sexual abuse material (Art. 5(1), points (ba) and (bb)).
- 2 December 2027: The obligations for high-risk AI systems under Annex III become applicable, for example in the areas of HR, education or credit scoring. This is the central date for companies that deploy or develop AI in such sensitive areas.
- 2 August 2028: The obligations for high-risk AI systems under Annex I become applicable, i.e. for AI in products covered by the EU harmonisation legislation listed in Annex I, Section A (e.g. medical devices).
What applies today?
Since 2 February 2025, the prohibitions under Article 5 and the AI literacy obligation under Article 4 have applied, to virtually every company that uses AI professionally. Since 2 August 2026 the Regulation has applied generally. Two new prohibitions follow on 2 December 2026. The postponed high-risk obligations apply only from 2 December 2027 (Annex III) and 2 August 2028 (Annex I) respectively.
Prohibited practices (Art. 5): Certain AI applications have since been fully banned. These include systems that manipulate people without their knowledge, social scoring and real-time remote biometric identification in publicly accessible spaces for law enforcement purposes, apart from narrowly defined exceptions. Anyone who places such systems on the market, puts them into service or uses them has been infringing Article 5 since February 2025.
AI literacy obligation (Art. 4): Alongside this, Article 4 applies. In its current version it requires providers and deployers to take measures to support the development of the AI literacy of their staff, taking account of knowledge, experience and context of use. This obligation explicitly covers deployers too, i.e. companies that use finished AI products such as chatbots, analysis tools or AI-powered recruitment software without developing them themselves.
What comes next?
Next, on 2 December 2026, two new prohibitions take effect and the transitional period for labelling under Art. 50(2) ends. The high-risk obligations follow. Under the Digital Omnibus (Regulation (EU) 2026/1744, in force since 27 July 2026), the core provisions no longer apply from the originally planned dates (2 August 2026 for Annex III, 2 August 2027 for Annex I), but from 2 December 2027 for high-risk AI under Annex III and from 2 August 2028 for high-risk AI under Annex I. This gives companies more time to prepare, but they should use it for an inventory, risk classification and documentation.
What counts as high-risk? The EU AI Act defines specific areas of application in Annex III: AI in recruitment and HR management, in educational and vocational training institutions, in credit scoring, in critical infrastructure and in law enforcement. AI systems that prepare or influence decisions about access to essential public or private services can also be classified as high-risk.
For high-risk systems, from the dates mentioned above, the requirements include, among others: a documented risk management system, technical documentation, human oversight and a quality management system. The AI literacy of the persons deployed under Article 4 applies independently of this, already today. The risk-based approach and the risk levels are explained on our EU AI Act overview page.
What companies should do and by when
- Immediately: create an AI inventory: Where does your company use AI systems, and which risk class do those applications belong to?
- Immediately (if not already done): organise AI literacy training for all employees who use AI professionally, and document participation. This makes it easier to demonstrate the measures later on.
- Before 2 December 2027 (Annex III) or 2 August 2028 (Annex I): for high-risk AI, check whether risk management, technical documentation and human oversight can be put in place in accordance with the EU AI Act's requirements.
- Ongoing: when introducing new AI tools and GPAI models, assess their risk class and any transparency obligations before putting them into production use.
- Ongoing: keep training records safely and refresh them regularly. AI is evolving fast; regular refreshers keep knowledge up to date. Art. 4 sets no fixed deadline for this.
Frequently asked questions about the deadlines
The AI literacy obligation already applies, but what exactly does that mean?
Article 4 of the EU AI Act requires providers and deployers to take measures to support the development of the AI literacy of their staff and of other persons dealing with the operation and use of AI systems on their behalf. Since 27 July 2026, the reworded version under the Digital Omnibus has applied: no one has to guarantee a particular level of competence. Art. 4 sets out criteria, but no learning catalogue. It is not about all employees being able to code. In practice, they should understand how the systems in use work, what opportunities and risks they carry and when they need to check or correct outputs. A structured EU AI Act training course is a simple, documentable measure for meeting this obligation. More details are explained on the Art. 4: AI literacy obligation page.
We only use simple AI tools. Does this really affect us?
Yes. The AI literacy obligation under Article 4 applies regardless of whether a company develops AI or "merely" uses it. Anyone who professionally uses an AI chatbot, an AI-powered translation tool or automated analysis software under their own authority is a deployer within the meaning of the EU AI Act (Art. 3(4)) and therefore bound by the obligation. The risk class of the AI in use determines the scope of further obligations. But the literacy obligation applies to all. Further answers can be found on our FAQ page.
Train your team now Understand the AI literacy obligation
This overview is based on the application timeline (Art. 113) of Regulation (EU) 2024/1689 as amended by Regulation (EU) 2026/1744 and does not constitute legal advice.