AI Act: Regulation (EU) 2024/1689
CELEX 32024R1689 · Read 2026-08-12
The act behind our transparency pages: what a reader must be told about a rule that produces a number, and how that rule is documented.
Current version
- Last verified
- 2026-08-31
- Source
- CELEX 32024R1689
- Consolidated version
- 2025-11-20(derived)
- Amendments
- Regulation (EU) 2025/384 (Digital Omnibus on AI) införd 2025-11-20
- Status
- In force · updated · consolidated
Chain of evidence
- Retrieved from EUR-Lex
- Verified against the consolidated version
- Automatic check on every EU amendment
This act is updated automatically when the EU publishes a new version.
Download the record
The same fields as above, for audit and archiving.
AI-förordningen för AI-system
Bygger du ett AI-system som behöver AI-Act-kompatibel rättskälla? Vi levererar officiell lydelse, daglig färskhet och full proveniens via API, utan att du behöver lägga crawl-budget.
Identifier
- CELEX
- 32024R1689
- Identifier
- EU 2024/1689
- Adopted
- 2024-06-13
- In force
- 2024-08-01
- Applies from
- 2026-08-02
- Articles
- 113
Chapters
- I General provisions1-4
- II Prohibited AI practices5
- III High-risk AI systems6-49
- IV Transparency obligations for providers and deployers of certain AI systems50
- V General-purpose AI models51-56
- VI Measures in support of innovation57-63
- VII Governance64-70
- VIII EU database for high-risk AI systems71
- IX Post-market monitoring, information sharing and market surveillance72-94
- X Codes of conduct and guidelines95-96
- XI Delegation of power and committee procedure97-98
- XII Penalties99-101
- XIII Final provisions102-113
Articles
113 articles are read into the register, one address each. Read on 2026-08-18.
Recitals that carry the interpretation
The recitals in the preamble have no legal force of their own but often decide how the article is read. The links are curated against the official text. Read 2026-08-31.
Skäl 27 · Article 6
The high-risk classification: product safety legislation and Annex III as the two tracks.
Skäl 28 · Article 5
Why manipulation and exploitation of vulnerability are banned: practices that undermine free choice.
Skäl 44 · Article 5
The carve-outs for law-enforcement biometric identification: strict proportionality and judicial review.
Skäl 89 · Article 9
The risk-management system as a continuous process over the system's whole lifecycle.
Skäl 97 · Article 17
Quality management systems for providers: documentation, allocation of responsibility and compliance control.
Skäl 105 · Article 43
Conformity assessment: for which high-risk systems self-assessment suffices and when a notified body is required.
Skäl 109 · Article 49
Registration in the EU database: what information must be given and who has access.
Skäl 132 · Article 50
The transparency duties: the user must know when it interacts with an AI system or synthetic content.
Skäl 109 · Article 71
The EU database's function as the public register of high-risk systems.
Skäl 165 · Article 99
The penalties: turnover-based fines must be dissuasive even for the largest actors.
Guidance and secondary law
Authority guidance and the delegated or implementing acts that fill the article, with the primary source at the issuing body. Read 2026-08-31.
- Guideline · Europeiska kommissionen2025-07-18Guidelines on the scope of obligations for general-purpose AI models
Which obligations bind whom under Chapter V: the provider concept, downstream modification and the open-source carve-outs.
Article 51, 52, 53, 54, 55
- Guideline · Europeiska kommissionen2025-07-10General-Purpose AI Code of Practice
The voluntary code for general-purpose models: transparency, copyright and systemic risk, read against Articles 53 and 55.
Article 53, 55
- Guideline · Europeiska kommissionen2025-02-04Commission Guidelines on prohibited artificial intelligence practices under Regulation (EU) 2024/1689
The Commission's reading of the prohibited AI practices in Article 5: manipulation, vulnerability, social scoring and biometric categorisation.
Article 5
Amendment history
The amending acts that affect the base act, so the question of what applied on a given date can be answered against the register. Read 2026-08-31.
Postpones the application of the Chapter III high-risk regime and ties it to the readiness of standards.
CELEX 32025R0384
Supervisory bodies
Official source
National instruments
DK Lov om supplerende bestemmelser til forordningen om kunstig intelligens
LOV nr 467 af 14/05/2025 · www.retsinformation.dk
FI Laki eräiden tekoälyjärjestelmien valvonnasta
1377/2025 · www.finlex.fi
DE Gesetz zur Durchführung der Verordnung (EU) 2024/1689 (KI-MIG)
KI-MIG · www.gesetze-im-internet.de
SE Svensk författningssamling, Riksdagen
Not stated in the register · www.riksdagen.se
NO Lovdata
Not stated in the register · lovdata.no
UK legislation.gov.uk
Not stated in the register · www.legislation.gov.uk
Obligations, actions and sanctions
Article 16 · Role: provider
Meet the provider obligations for high-risk AI systems.
Deadline: Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 9 · Role: provider
Establish and maintain a risk management system.
Deadline: Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 10 · Role: provider
Apply data and data governance requirements to training, validation and testing data.
Deadline: Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 11 · Role: provider
Draw up and keep technical documentation.
Deadline: Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 12 · Role: provider
Enable automatic recording of events (logging).
Deadline: Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 17 · Role: provider
Operate a quality management system.
Deadline: Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 43 · Role: provider
Carry out the applicable conformity assessment before placing on the market.
Deadline: Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 72 · Role: provider
Run post-market monitoring of the system in use.
Deadline: Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 73 · Role: provider
Report serious incidents to the market surveillance authority.
Deadline: Serious incident: not later than 15 days after awareness (Art. 73(2)); not later than 2 days for a widespread infringement or a serious and irreversible disruption of critical infrastructure (Art. 73(3), Art. 3(49)(b)); not later than 10 days in the event of the death of a person (Art. 73(4)).
Decision chain (JSON)Article 26 · Role: deployer
Use the system in line with the instructions and assign human oversight.
Deadline: Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 27 · Role: deployer
Carry out a fundamental rights impact assessment where required.
Deadline: Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 23 · Role: importer
Verify conformity, documentation and marking before placing on the market.
Deadline: Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 24 · Role: distributor
Verify marking and documentation before making available on the market.
Deadline: Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 53 · Role: gpai-provider
Keep model documentation and a copyright policy for general-purpose AI models.
Deadline: Applies since 2 August 2025; models placed on the market before that date must comply by 2 August 2027. Commission fines apply from 2 August 2026.
Decision chain (JSON)Article 55 · Role: gpai-provider
Meet the additional obligations for models with systemic risk.
Deadline: Applies since 2 August 2025 for models with systemic risk; notify the Commission within two weeks of meeting the threshold (Art. 52(1)). Models placed on the market before that date must comply by 2 August 2027.
Decision chain (JSON)Article 5 · Role: operator
Do not place on the market, put into service or use AI systems for the prohibited practices in Article 5, including manipulative techniques, exploitation of vulnerabilities, social scoring, untargeted scraping of facial images, emotion recognition at work or in education, and biometric categorisation by sensitive traits.
Deadline: Applies since 2 February 2025.
Decision chain (JSON)Article 4 · Role: provider-or-deployer
Take measures to ensure a sufficient level of AI literacy among staff and others operating or using AI systems on their behalf, taking account of their knowledge, experience, education and context of use.
Deadline: Applies since 2 February 2025; amended wording applies from 27 July 2026. No dedicated fine in Article 99; enforcement follows national law.
Decision chain (JSON)Article 50 · Role: provider
Design AI systems that interact with natural persons so they are informed they are interacting with an AI system, and mark synthetic audio, image, video and text output in a machine-readable and detectable format.
Deadline: Applies since 2 August 2026; providers of generative systems placed on the market before that date have until 2 December 2026 for the marking in Article 50(2).
Decision chain (JSON)Article 50 · Role: deployer
Inform natural persons exposed to an emotion recognition or biometric categorisation system, and disclose that deep fake content or AI-generated text published to inform the public on matters of public interest has been artificially generated or manipulated.
Deadline: Applies since 2 August 2026, at the latest at the time of the first interaction or exposure.
Decision chain (JSON)Article 54 · Role: gpai-provider
Appoint, by written mandate, an authorised representative established in the Union before placing a general-purpose AI model on the Union market when established outside the Union.
Deadline: Applies since 2 August 2025; models placed on the market before that date must comply by 2 August 2027. Commission fines apply from 2 August 2026.
Decision chain (JSON)Article 13 · Role: provider
Design high-risk AI systems to be sufficiently transparent and accompany them with instructions for use stating the provider's identity, the system's characteristics, capabilities, limitations and human oversight measures.
Deadline: Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 14 · Role: provider
Design high-risk AI systems so that natural persons can effectively oversee them during use, including the ability to disregard, override or stop the system.
Deadline: Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 15 · Role: provider
Design high-risk AI systems to achieve an appropriate level of accuracy, robustness and cybersecurity throughout their lifecycle, and declare accuracy metrics in the instructions for use.
Deadline: Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 18 · Role: provider
Keep the technical documentation, quality management system documentation, notified body decisions and EU declaration of conformity at the disposal of national competent authorities.
Deadline: Ten years after the system has been placed on the market or put into service. Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 19 · Role: provider
Keep the logs automatically generated by the high-risk AI system, to the extent they are under the provider's control.
Deadline: For a period appropriate to the intended purpose, of at least six months. Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 20 · Role: provider
Immediately take corrective action to bring a non-conforming high-risk AI system into conformity, withdraw, disable or recall it, and inform distributors, deployers and the authorities of a risk.
Deadline: Immediately on considering or having reason to consider non-conformity. Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 47 · Role: provider
Draw up a written, machine-readable EU declaration of conformity for each high-risk AI system and keep it up to date.
Deadline: Keep for ten years after placing on the market or putting into service. Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 48 · Role: provider
Affix the CE marking visibly, legibly and indelibly to the high-risk AI system, or digitally for systems provided digitally.
Deadline: Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 49 · Role: provider
Register the provider and the high-risk AI system listed in Annex III in the EU database before placing it on the market or putting it into service.
Deadline: Before placing on the market or putting into service. Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 26(6) · Role: deployer
Keep the logs automatically generated by the high-risk AI system, to the extent under the deployer's control, and inform workers' representatives and affected workers before putting the system into service at the workplace.
Deadline: Logs for a period appropriate to the intended purpose, of at least six months. Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)Article 86 · Role: deployer
Provide affected persons, on request, with a clear and meaningful explanation of the role of the high-risk AI system in a decision that produces legal or similarly significant effects.
Deadline: Applies from 2 December 2027 for high-risk systems under Article 6(2) and Annex III, and from 2 August 2028 under Article 6(1) and Annex I (Article 113 as amended by Regulation (EU) 2026/1744).
Decision chain (JSON)
Sanction: Administrative fine: prohibited practices up to EUR 35 million or 7 % of worldwide annual turnover; other obligations up to EUR 15 million or 3 %; incorrect information up to EUR 7.5 million or 1 %. For general-purpose AI models the Commission (AI Office) decides, up to EUR 15 million or 3 %.
Summary of the obligation in the register; the official language versions on EUR-Lex are authoritative. Source: NovaCopilot
Change monitoring
The row is re-read against the publisher's own publication on a fixed interval. The dates below can be checked on the spot and travel with the API response.
| Field | Value |
|---|---|
| Register | EUR-Lex |
| Publisher | Europeiska unionens publikationsbyrå |
| Last read | 2026-08-31 |
| Interval | Every 30 days |
| Next re-read | 2026-09-30 |
| Status | Re-read scheduled |
Row history
Dated events concerning this exact row, newest first. No event appears here without a date in a primary source.
| Date | Event | Source |
|---|---|---|
| 2026-08-31 | Row read against EUR-Lex | Europeiska unionens publikationsbyrå |
| 2026-08-02 | Date of application under the act | EUR-Lex |
| 2024-08-01 | Entry into force under CELEX 32024R1689 | EUR-Lex |
Monitoring states when the row was checked, not how the legal position should be assessed.
Cite this row
The string below travels with a memo, a case file or an agent chain. The same string sits in the citation field of the API response.
CELEX 32024R1689, AI Act: Regulation (EU) 2024/1689. NovaCopilot, https://legal.exploreworldai.com/eu/rattsakter/ai-act-2024-1689 (hämtad 2026-10-25, bevis sha256:0a00d6c14a5f39d7, bygge legal-2026-10-25).
- sha256:
- 0a00d6c14a5f39d79c67a37370c3cfea9a68e56feb8361644bde41a7d5be967d
- hämtad:
- 2026-10-25
- source_confidence:
- official
- bygge:
- legal-2026-10-25
Official publication. The address points to the authority or court that published the row. Europeiska unionens publikationsbyrå, EUR-Lex.
The row is a source reference with an official identifier. No legal advice and no compliance decision.
Verifiable trust signals
- Source-linked answers and explicit evidence gaps
- Obligations, controls, risks and consequences
- Stable objects and verifiable checksums
- Rule-based objects, optional source-linked reasoning
- Model card, source status and audit