EU AI Act: deadlines and news
The AI Act sets rules for AI systems and general-purpose AI models in the EU. Regulation (EU) 2026/1744, in force since 27 July 2026, moved the high-risk dates to 2 December 2027 and 2 August 2028. Every entry names its source and says whether it is binding law or guidance.
Coming up
Deadline for Article 50(2) marking by generative AI systems already on the market
Providers of AI systems, including general-purpose AI systems, that generate synthetic audio, image, video or text and were placed on the market before 2 August 2026 must take the necessary steps to comply with Article 50(2) by this date (Art. 111(4), added by Regulation (EU) 2026/1744).
Prohibitions on AI generating intimate images and child sexual abuse material apply
Article 5(1), first subparagraph, points (ba) and (bb), and Article 5(1a) and (1b), added by Regulation (EU) 2026/1744, apply from this date (Art. 113(3), point (a), as amended). They cover AI that generates or manipulates intimate or sexually explicit material of identifiable people without their explicit consent, and child sexual abuse material.
Deadline for Commission guidelines on the AI Act and Annex I product legislation
Article 96(1), point (g), added by Regulation (EU) 2026/1744, requires the Commission to publish guidelines on applying Articles 8(2), 9(10) and 17(3) alongside Annex I Section A product legislation by this date. The guidelines will be non-binding.
Deadline for Commission delegated acts limiting overlapping high-risk requirements
Article 2(13), added by Regulation (EU) 2026/1744, allows certain Articles 9 to 15 and 17 to 25 requirements to be limited where Annex I Section A legislation gives equal protection. The Commission must adopt delegated acts specifying the details by this date.
Deadline for general-purpose AI models placed on the market before 2 August 2025
Providers of general-purpose AI models placed on the market before 2 August 2025 must take the necessary steps to comply with the Regulation by this date (Art. 111(3)).
National AI regulatory sandboxes must be operational
Member States must ensure that their competent authorities establish at least one national AI regulatory sandbox that is operational by this date (Art. 57(1), as amended by Regulation (EU) 2026/1744). The original date was 2 August 2026.
Deadline for Commission guidance and template on post-market monitoring plans
Article 72(3), as amended by Regulation (EU) 2026/1744, no longer provides for an implementing act with a template. Instead the Commission must adopt guidance including a template on the post-market monitoring plan by this date.
Rules for Annex III high-risk AI systems apply
Chapter III Sections 1, 2 and 3 apply to AI systems classified as high-risk under Article 6(2) and Annex III from this date (Art. 113(3), point (c)(i), as amended by Regulation (EU) 2026/1744). The original date was 2 August 2026, and the new date is fixed in the text.
Rules for high-risk AI systems in Annex I products apply
Chapter III Sections 1, 2 and 3 apply to AI systems classified as high-risk under Article 6(1) and Annex I from this date (Art. 113(3), point (c)(ii), as amended by Regulation (EU) 2026/1744). Article 6(1) and its obligations were originally due to apply from 2 August 2027.
Deadline for high-risk AI systems intended for public authorities placed on the market earlier
Providers and deployers of high-risk AI systems intended for use by public authorities that were placed on the market before the Chapter III date of application must take the necessary steps to comply by this date (Art. 111(2), as amended by Regulation (EU) 2026/1744).
Deadline for AI components of large-scale EU IT systems placed on the market earlier
AI systems that are components of the large-scale IT systems listed in Annex X and were placed on the market or put into service before 2 August 2027 must be brought into compliance by this date (Art. 111(1)).
2026
AI Office publishes a report on requirements for external evaluators of GPAI models
The report follows a workshop on 15 July 2026 on the qualification and independence of external evaluators of general-purpose AI models with systemic risk. It states the current views of the AI Office, which may change, and is not binding.
First harmonised standard for the AI Act referenced in the Official Journal
Implementing Decision (EU) 2026/2274 of 12 October 2026 references EN 18286:2026 on AI quality management systems, and the Commission announced it on 19 October. Providers who apply it are presumed to conform with the Article 17 requirements it covers (Art. 40), and using it is voluntary.
AI Scientific Panel holds a special meeting on loss-of-control incidents
The Commission reports that the AI Office has requested and received serious incident reports and sent requests for information on security practices, and that it will send further requests. The panel's recommendations are advice, not binding.
Italy: Legislative Decree 179/2026 sets the powers of the national AI authorities and the penalties
Legislative Decree No. 179 of 7 October 2026 (Gazzetta Ufficiale n. 234 of 8 October 2026) names AgID as notifying authority and ACN as market surveillance authority and single point of contact, with the Bank of Italy, CONSOB, IVASS and the data protection authority for their areas, and sets sanctions and rules on AI in education. The Council of Ministers gave final approval on 4 August 2026. It is enacted law and enters into force on 23 October 2026.
Spain: the AI governance bill lapses as the Cortes are dissolved
Royal Decree 806/2026, published in the BOE on 6 October 2026, dissolved the Congress and the Senate and called general elections for 29 November 2026. The Congress file for bill 121/000096 shows it as concluded and lapsed from 6 October 2026, so the bill cannot pass in this legislature and a new government would have to table a fresh one.
Poland: Sejm and Senate appoint the first chair of the KRiBSI
The Sejm voted on 18 September 2026 to appoint Pamela Krzypkowska as chair of the KRiBSI (271 for, 24 against, 140 abstentions) and the Senate gave its consent on 24 September; both resolutions appeared in Monitor Polski on 25 September (M.P. 2026 poz. 945 and 946). The chair now leads the authority that the enacted AI Act made the market surveillance authority.
Slovenia: AKOS explains the labelling duties of providers of generative AI systems
AKOS publishes an explainer on Article 50(2): providers of AI systems that create or alter synthetic audio, image, video or text must mark the output from 2 August 2026, with a transition until 2 December 2026 for systems placed on the market earlier. It is authority guidance, not law.
Italy: Legislative Decree 160/2026 on AI in police work and civil and criminal liability is published
Legislative Decree No. 160 of 9 September 2026, published in Gazzetta Ufficiale n. 214 of 15 September 2026, adapts Italian law to the AI Act on the use of AI systems for police activity and on civil and criminal liability, under the delegation in Article 24 of Law 132/2025. It is enacted law and entered into force on 30 September 2026.
Slovakia: Government approves the bill on state administration in artificial intelligence
By resolution 349/2026 of 26 August 2026 the government approved the MIRRI bill that implements the AI Act, with one remark adopted in the meeting, and sent it to parliament. MIRRI becomes general market surveillance authority and single point of contact, the standards office UNMS is notifying authority and the Ombudsman gets fundamental rights powers; it applies from 1 January 2027 and is a bill, not law.
Slovenia: AKOS publishes Slovenian icons for labelling AI-generated content
AKOS, the national entry point for the AI Act, publishes the Slovenian icons for AI generated and AI altered content for the labelling duties of Article 50, which apply from 2 August 2026. The duties are binding, use of the icons is voluntary, and the page is authority guidance, not law.
Cyprus: public consultation opens on the draft Artificial Intelligence Law of 2026
The e-Consultation portal shows a public consultation on the draft Artificial Intelligence Law of 2026, open from 16 August to 16 September 2026, to implement Regulation (EU) 2024/1689. According to the Commissioner of Communications, the draft sets the national governance, market surveillance powers, sanctions and an AI regulatory sandbox. It is a draft, not enacted law.
Czech Republic: Legislative Council interrupts discussion of the AI bill
At its 336th meeting on 6 August 2026 the Legislative Council of the Government interrupted its discussion of the Minister of Industry and Trade's bill on artificial intelligence (ref. 503/26). The government has therefore not yet approved the bill and it is not law.
General date of application of the AI Act
The Regulation applies from this date unless a special date applies (Art. 113, second paragraph), including the Article 50 transparency duties and the Commission's fining power over general-purpose AI providers (Art. 101). The Chapter III high-risk rules no longer start on this date after Regulation (EU) 2026/1744.
About 190 organisations sign the Code of Practice on Transparency of AI-generated Content
The Commission reported about 190 signatories by the end of July 2026. Its list, checked on 21 October 2026, shows 95 signatories of the provider section and 192 of the deployer section; signing is voluntary.
Ireland: Regulation of Artificial Intelligence Act 2026 comes into operation
S.I. No. 403 of 2026 appoints 31 July 2026 as the day the Act comes into operation, except two paragraphs of the definition of 'applicable market surveillance authority' in section 78, which still need a separate order. The establishment day of the AI Office is appointed separately under section 7.
Germany: KI-MIG enters into force and the Bundesnetzagentur becomes the AI market surveillance authority
The KI-Marktueberwachungs- und Innovationsfoerderungs-Gesetz of 22 July 2026 (BGBl. 2026 I Nr. 223, published 28 July) applies from 29 July 2026. It makes the Bundesnetzagentur the central market surveillance authority, contact point and complaints office, while sector authorities keep their own areas.
Digital Omnibus on AI enters into force
Regulation (EU) 2026/1744 is adopted law, not a proposal. It moves the high-risk application dates to 2 December 2027 and 2 August 2028, makes Article 4 a duty to support AI literacy, and applies Articles 102 to 110 of the AI Act from this day (Art. 113(3), point (d)).
Poland: Act on artificial intelligence systems is published and designates the KRiBSI
The Act of 3 July 2026 (Dz.U. 2026 poz. 1003) is enacted law. It makes the new Commission for the Development and Safety of AI (KRiBSI) the market surveillance authority and single point of contact, names the digitalisation minister as notifying authority and gives the KRiBSI the national regulatory sandbox. Most of it applies from 11 August 2026; inspections, proceedings, fines and related chapters follow on 28 October 2026.
Digital Omnibus on AI published in the Official Journal as Regulation (EU) 2026/1744
Regulation (EU) 2026/1744 of 8 July 2026 amends Regulations (EU) 2024/1689, (EU) 2018/1139 and (EU) 2023/1230. It is binding law and entered into force on the third day after publication.
Romania: ANCOM sets out where AI Act implementation stands and its proposed role
A press release of 24 July 2026 says a Government Memorandum proposes ANCOM as national market surveillance authority and single point of contact, alongside sectoral authorities, and that a national law with the authorities' powers and the sanctions procedure is still being drafted. This is an authority statement on a government memorandum, not enacted law.
Ireland: Regulation of Artificial Intelligence Act 2026 signed into law
Act No. 31 of 2026, signed on 21 July 2026 after passing both Houses, creates the AI Office of Ireland as single point of contact, provides for the regulatory sandbox, lists the Article 77 fundamental rights bodies, and amends the Central Bank, communications regulation, consumer protection and freedom of information statutes. It is enacted law and comes into operation by ministerial order.
Commission publishes guidelines on the Article 50 transparency obligations
The guidelines cover scope, definitions, the transparency duties for providers and deployers, and exceptions, ahead of their application on 2 August 2026. They are non-binding guidance.
Greece: Law 5321/2026 sets the national framework for the AI Act
Law 5321/2026 (Government Gazette A 114 of 20 July 2026) names the Hellenic Data Protection Authority as market surveillance authority for prohibited practices, Annex III high-risk systems and Article 50 transparency, and as single point of contact. The telecommunications regulator EETT becomes notifying authority and hosts the AI sandbox together with the data protection authority. It is enacted law; the digital ministry announced it on 22 July 2026.
Germany: Bundesrat lets the AI Act implementation law pass
The Bundesrat did not call the mediation committee on 10 July 2026, so the law could be signed and promulgated.
Luxembourg Council of State gives its opinion on the AI Act bill
The Council of State adopts its opinion on bill 8476 on 10 July 2026, asking for changes to some provisions under threat of formal objection. The bill is still in committee in the Chamber of Deputies and is not law.
Commission and AI Board assess the transparency code of practice as adequate
On 8 July the Commission concluded that the code adequately covers Article 50(2), (4) and (5), and the AI Board adopted its assessment on 9 July. Adherence is not conclusive evidence of compliance.
Council adopts the Digital Omnibus on AI
The Council's decision completed adoption of the act. It was signed on 8 July 2026 and published on 24 July 2026.
Parliament approves the agreed Digital Omnibus on AI at first reading
Parliament adopted its first-reading position (T10-0198/2026) on the agreed text. The act was not yet law, because Council adoption was still missing.
Portugal: ANACOM consults on draft recommendations for the AI Act ban on prohibited practices
On 16 June 2026 ANACOM decided to open a public consultation on draft recommendations for implementing Article 5 of the AI Act, with comments due by 16 July 2026. The draft is non-binding and sets out organisational recommendations and procedures; the legal duties come from the Regulation itself.
Spain: Congress publishes the AI governance bill and opens the amendment period
The Official Bulletin of the Cortes Generales of 12 June 2026 published bill 121/000096, referred it to the Economy, Trade and Digital Transformation Committee and set 30 June 2026 as the deadline for amendments. The Congress file then lists weekly extensions up to 7 October 2026. It was a bill in committee stage, not law.
Germany: Bundestag passes the AI Act implementation law
The Bundestag adopted the bill on 11 June 2026 in the version recommended by the Digital Committee. The law still needed the Bundesrat and publication.
Poland: Sejm passes the bill on artificial intelligence systems that creates the KRiBSI
On 11 June 2026 the Sejm passed the government bill on artificial intelligence systems in third reading. The Senate proposed amendments (25 June) and the Sejm dealt with them on 3 July, the date the final Act carries; at this step it was still a bill, not law.
Commission publishes the final Code of Practice on marking and labelling AI-generated content
The voluntary code helps providers and deployers of generative AI meet the Article 50 duties that apply from 2 August 2026. It is non-binding, and Article 50 itself is a legal obligation.
Sweden: Government assigns five agencies to act as national AI Act authorities until the end of 2026
In a decision of 4 June 2026 (published on 12 June) the Government gave PTS, the Authority for Privacy Protection (IMY), Finansinspektionen, the Medical Products Agency and Swedac the task of acting as national competent authorities, with PTS as single point of contact and operator of the sandbox. It is a Government instruction that runs to 31 December 2026 while the complementary law is pending, not a statute.
Commission appoints the AI Scientific Panel and the Advisory Forum
The Scientific Panel has 60 independent experts who advise the AI Office and national authorities, and the Advisory Forum advises on implementation and standardisation. Members serve two-year terms.
Spain: Council of Ministers sends the AI governance bill to Parliament
On 26 May 2026 the Council of Ministers approved the bill for an Organic Law on the good use and governance of artificial intelligence and sent it to the Congress of Deputies. The bill would make AESIA the main supervisor and single point of contact, keep the existing sector authorities for product law, set fines of up to EUR 35 million or 7 percent of turnover and give the national sandbox to AESIA. It was a bill, not law.
Commission publishes its report on the review of prohibitions and high-risk use cases
The annual review under Article 112(1) notes a possible gap for AI generating child sexual abuse material and non-consensual intimate content, and says a fuller assessment of Article 5 needs at least a year of application. It does not change the law by itself.
Commission publishes draft guidelines on classifying AI systems as high-risk
The draft explains Article 6 with practical examples and was open to a targeted consultation until 23 July 2026, after Article 6(5) had set 2 February 2026 for the guidelines. It is a non-binding draft, and the Commission page updated on 19 October 2026 still shows no final version.
Parliament and Council reach a provisional agreement on the Digital Omnibus on AI
Negotiators agreed fixed high-risk dates of 2 December 2027 (Annex III) and 2 August 2028 (Annex I products) and a ban on AI generating non-consensual intimate content and child sexual abuse material. A provisional agreement is not law until both institutions formally adopt it.
Netherlands opens consultation on its AI Regulation implementation act
On 20 April 2026 State Secretary Aerdts puts the draft implementation act for the AI Regulation into internet consultation until 1 June 2026. Existing supervisors each oversee AI in their own domain, the Autoriteit Persoonsgegevens covers areas without a supervisor and shares a coordinating role with the RDI; this is a draft, not law.
Parliament adopts its position on the Digital Omnibus on AI
Parliament adopted its first-reading position (T10-0098/2026) after a joint IMCO and LIBE committee report of 18 March, and sent the file back for negotiations with the Council. A negotiating position is not law.
Germany: AI sandbox pilot with Hesse and the data protection commissioner ends
The Bundesnetzagentur, the Hessian digital ministry and the Federal Data Protection Commissioner publish the final report of a pilot that simulated an AI regulatory sandbox with two start-up cases, including a roadmap for medical devices with high-risk AI. The report is non-binding and prepares the Bundesnetzagentur's real sandbox.
Council agrees its position on the Digital Omnibus on AI
The Council agreed its negotiating mandate: fixed high-risk dates of 2 December 2027 and 2 August 2028 instead of dates tied to standards, plus a new ban on AI generating non-consensual intimate content and child sexual abuse material. A mandate is not law.
Germany: Federal Government submits the AI Act implementation bill to the Bundestag
Bundestag document 21/4594 of 9 March 2026 contains the government bill, with the Bundesnetzagentur as central market surveillance authority. A bill is not law.
Croatia: parliament's 2026 alignment plan lists the AI Act implementation law
On 6 March 2026 the Sabor adopted the 2026 plan for aligning Croatian law with EU law (NN 26/2026), which schedules an act implementing Regulation (EU) 2024/1689 for the first quarter under the Ministry of Justice, Public Administration and Digital Transformation. A full text search of Narodne novine on 21 October 2026 found no such act, so it is planned and not yet law.
French Senate adopts EU adaptation bill with first AI Act provisions
On 18 February 2026 the Senate adopts, at first reading, the bill adapting French law to EU law. Its Article 24 adds a title on AI systems to the 2004 digital economy law and Article 24 bis authorises an ordinance on regulatory sandboxes; the text went to the National Assembly on 20 February 2026 and is a bill, not law.
Denmark: bill for full AI Act supervision introduced in February 2026 but lapsed
Bill L 111 (AI-loven), introduced on 18 February 2026 and read for the first time on 17 March 2026, would have replaced Law 467/2025 and added supervision of high-risk systems and transparency duties from 2 August 2026. Folketingstidende lists it as lapsed (bortfaldet), so Law 467/2025 on prohibited practices remains the enacted Danish law.
Finland: law on the supervision of certain AI systems enters into force
Act 1377/2025, published on 29 December 2025, names Traficom as single point of contact and supervisor of the transparency duties, the Data Protection Ombudsman for prohibited practices, and sector authorities for high-risk systems. It is enacted Finnish law in force from 1 January 2026; the regulatory sandbox is left to a second act that went to consultation as a draft in 2025.
2025
Spain: AESIA publishes 16 non-binding guides on complying with the AI Act
On 16 December 2025 the Spanish AI supervision agency AESIA published 16 practical guides for high-risk AI systems, the result of Spain's AI regulatory sandbox. The guides are not binding and do not replace the AI Act; AESIA said they would be updated once the Digital Omnibus on AI is approved.
Commission launches the AI Act whistleblower tool
The tool gives individuals a confidential channel to report suspected AI Act breaches to the AI Office, in any EU official language.
Slovakia: Act 318/2025 adds fines for authorised conformity assessment bodies under the AI Act
Act No. 318/2025 Coll., adopted on 21 October 2025 and published on 21 November 2025, amends the conformity assessment Act 56/2018. From 1 January 2026 the standards office can fine authorised bodies at the Art. 99(4) and (6) AI Act amounts for breaching the Art. 99(4)(f) duties; it is enacted law but does not designate the AI market surveillance authority.
Commission proposes the Digital Omnibus on AI
Proposal COM(2025) 836 would amend the AI Act and Regulation (EU) 2018/1139, including a delay of the high-risk rules tied to the availability of standards with long-stop dates of 2 December 2027 and 2 August 2028. A proposal is not law.
Commission publishes a template for reporting serious incidents of systemic risk GPAI models
The template supports the Article 55 duty to report serious incidents to the AI Office and helps providers show compliance with Commitment 9 of the GPAI Code of Practice. It is a non-binding Commission document.
Hungary: Act LXXV of 2025 and Government Decree 344/2025 implement the AI Act
Published in Magyar Kozlony on 31 October 2025, Act LXXV of 2025 is enacted law that sets up the AI notifying and market surveillance authorities, a regulatory sandbox and a Hungarian AI Council. Government Decree 344/2025 designates the National Accreditation Authority as notifying authority and the minister responsible for enterprise development as market surveillance authority and single point of contact; both apply from 1 December 2025, the sandbox from 2 August 2026.
prEN 18286 on AI quality management systems enters public enquiry
The Commission names prEN 18286 as the first harmonised standard for AI to enter public enquiry. It is a draft standard at this stage and gives no presumption of conformity.
Slovenia: National Assembly adopts the act implementing the AI Act (ZIUDHPUI)
The National Assembly adopted the act implementing the AI Regulation (ZIUDHPUI), prepared by the Ministry of Digital Transformation, on 23 October 2025, and it was published on 6 November 2025. According to AKOS, the act makes the Agency for Communication Networks and Services the single contact point and gives it most supervisory powers and the regulatory sandbox; this is enacted Slovenian law.
Italy: Law 132/2025 enters into force and designates AgID and ACN as national AI authorities
Law No. 132 of 23 September 2025 (Gazzetta Ufficiale n. 223 of 25 September 2025) designates the Agency for Digital Italy (AgID) and the National Cybersecurity Agency (ACN) in Article 20. It is enacted Italian law, and the financial supervisors keep their market surveillance role.
Commission launches the AI Act Service Desk and Single Information Platform
The Single Information Platform offers an AI Act Explorer, a Compliance Checker and a form for questions to the Service Desk. It is an information service, not a legal instrument.
Sweden: inquiry proposes a complementary AI Act law with PTS as lead authority
The inquiry report SOU 2025:101, published on 6 October 2025, proposes a new complementary law and ordinance, eleven market surveillance authorities and two notifying authorities, with the Post and Telecom Authority (PTS) as lead authority and operator of the regulatory sandbox. It is a proposal, not law; the inquiry suggested entry into force on 2 August 2026.
Czech Republic: Ministry of Industry and Trade publishes its draft AI Act implementing law
A ministry press release of 26 September 2025 says the draft, then in inter-ministry comment, would make the Czech Telecommunication Office the single point of contact, with the Czech National Bank and the Data Protection Office as further supervisors, the Office for Technical Standardisation as notifying authority, the Ombudsman for fundamental rights and a national sandbox. It is a draft, not law.
Portugal: Government announces ANACOM as AI Act supervisory authority
ANACOM reports that at its conference on 19 September 2025 the Secretary of State for Digitalisation and the Infrastructure Minister announced that ANACOM will supervise the AI Act in Portugal. This was an announcement of intent, not a legal act: the Commission's list of single points of contact, last updated 7 September 2026, still marks the Portuguese designation as pending final adoption.
General-purpose AI, governance and penalty rules apply
Chapter V (general-purpose AI models), Chapter VII (governance), Chapter XII (penalties, except Article 101), Chapter III Section 4 (notified bodies) and Article 78 apply (Art. 113(3), point (b)). By this date Member States had to make the contact details of their competent authorities public (Art. 70(2)).
Denmark: law supplementing the AI Act applies and designates the competent authorities
Law No. 467 of 14 May 2025 designates the Agency for Digital Government (also the notifying authority), the Data Protection Authority and the Court Administration as national competent authorities, and applies from this date (section 12). It is enacted Danish law.
Commission and AI Board confirm the GPAI Code of Practice as adequate
They confirmed that the voluntary code is an adequate tool for general-purpose AI model providers to show compliance. The Commission lists 22 signatories on its code page (checked 21 October 2026), and xAI signed only the safety and security chapter.
Commission publishes the template for the public summary of GPAI training content
The template and explanatory notice set a common minimum baseline for the summary that Article 53(1), point (d), requires providers to publish. The Regulation requires use of the AI Office template.
Commission publishes guidelines on the scope of obligations for general-purpose AI model providers
The guidelines clarify who must comply with the general-purpose AI model obligations ahead of their application on 2 August 2025. They are non-binding guidance.
General-Purpose AI Code of Practice published
Independent experts delivered the voluntary code with chapters on transparency, copyright, and safety and security. It is non-binding, and signing is optional.
Commission replaces the AI standardisation request with Implementing Decision C(2025) 3871
The decision requests harmonised standards for high-risk AI systems that reflect the final AI Act, sets 31 August 2025 for drafting and expires on 28 February 2027. Regulation (EU) 2026/1744 later cites delayed preparation of standards as a reason to move the high-risk dates.
Commission adopts the implementing act for the AI scientific panel
The implementing act sets the rules for setting up and operating a panel of independent experts that advises the AI Office and national authorities, including alerts on risks from general-purpose AI models.
Latvia: Cabinet approves the report on implementing the AI Act
The Cabinet of Ministers approved the information report on implementing the AI Act at its meeting of 25 February 2025. The report names the Ministry of Economics as notifying authority, LATAK as accreditation body, VARAM as coordination contact point and the Ombudsman as fundamental rights authority; it is a government plan, not a law.
Commission publishes guidelines on the definition of an AI system
The guidelines explain how to apply the AI system definition in Article 3(1) to decide whether the Regulation covers a software system. They are non-binding and will be updated with practical experience.
AI Office publishes a living repository of AI literacy practices
The repository collects practices from AI Pact participants and other organisations on meeting Article 4. Copying a listed practice does not give a presumption of compliance with Article 4.
Commission publishes guidelines on prohibited AI practices
The guidelines explain the Article 5 prohibitions with legal reasoning and practical examples. They are non-binding, and only the Court of Justice gives authoritative interpretations.
Prohibited AI practices and AI literacy rules apply
Chapters I and II of the AI Act apply (Art. 113(3), point (a)): the Article 5 prohibitions and the Article 4 AI literacy duty. Regulation (EU) 2026/1744 later rewrote Article 4 and added two prohibitions that apply from 2 December 2026.
Lithuania: Seimas passes two laws that designate the national AI Act authorities
The Seimas adopted amendments to the Law on Technology and Innovation (XV-105) and the Law on Information Society Services (XV-106) on 14 January 2025, published on 22 January 2025. The Communications Regulatory Authority (RRT) is market surveillance authority and single contact point from 1 April 2025 and the Innovation Agency is notifying authority from 2 August 2025; this is enacted Lithuanian law.
2024
Luxembourg tables bill 8476 on AI Act authorities and sandbox
Bill 8476 is deposited in the Chamber of Deputies on 23 December 2024. It names ILNAS and the data protection commissioner's office as notifying authorities, the CNPD as default market surveillance authority with sector exceptions, and requires the CNPD to run a regulatory sandbox; it is a bill, not law.
Luxembourg government approves its AI Act implementation bill
On 29 November 2024 the Council of Government approves the bill implementing parts of Regulation (EU) 2024/1689 and naming the national authorities, with the data protection authority CNPD as single point of contact and default market surveillance authority. This is a bill, not law.
AI Act enters into force
Regulation (EU) 2024/1689 entered into force on the twentieth day after publication (Art. 113, first paragraph). Its rules apply in stages, starting with Chapters I and II on 2 February 2025.
AI Act published in the Official Journal
Regulation (EU) 2024/1689, signed on 13 June 2024, was published in the Official Journal on 12 July 2024. It is binding in its entirety and directly applicable in all Member States (Art. 113).
Council adopts the AI Act
The Council's final approval completed the adoption of the Regulation. It was signed on 13 June 2024 and published on 12 July 2024.
Parliament adopts the AI Act at first reading
Parliament endorsed the agreed text by 523 votes to 46 with 49 abstentions. The text still needed a lawyer-linguist check and formal Council approval.
Commission decision establishes the European AI Office
The Commission adopted a decision establishing the European AI Office within DG CONNECT. It is an internal Commission decision and does not affect the powers of national authorities.
2023
Council and Parliament reach a provisional agreement on the AI Act
Negotiators of the Council and Parliament agreed the text of the AI Act. A provisional agreement is not law until both institutions formally adopt it.
Parliament adopts its negotiating position on the AI Act
Parliament adopted its position by 499 votes to 28 with 93 abstentions and sent the file back for negotiations with the Council. A negotiating position is not law.
Commission requests CEN and CENELEC to draft European standards for AI
Implementing Decision C(2023) 3215 asked CEN and CENELEC to draft European standards for the key technical areas of the AI proposal, in support of Union AI policy, and they accepted. C(2025) 3871 later repealed it.
2022
Council adopts its general approach on the AI Act proposal
The Council adopted its position (general approach) on the proposal, which allowed it to negotiate with Parliament once Parliament had its own position. A general approach is a negotiating position, not law.
2021
CEN and CENELEC establish Joint Technical Committee 21 on artificial intelligence
JTC 21 develops European standards for AI, including harmonised standards in support of the AI Act. Applying such standards is voluntary, but a standard referenced in the Official Journal gives a presumption of conformity.
Commission proposes the Artificial Intelligence Act
The Commission published proposal COM(2021) 206 for a Regulation laying down harmonised rules on artificial intelligence, opening procedure 2021/0106(COD). A proposal is not law.
Sources
40
This page is regularly updated from the following official and journalistic sources.
Moet u zich met NIS2 bezighouden?
We begeleiden u stap voor stap. U vult in, wij leggen elk punt in eenvoudige woorden uit, en aan het eind tekent de directie.
