The EU AI Act's application dates changed in July 2026, and many compliance plans still work to the original high-risk deadline of 2 August 2026. Regulation (EU) 2026/1744, the Digital Omnibus on AI, moved the high-risk obligations for Annex III systems to 2 December 2027 and for Annex I product systems to 2 August 2028, added new prohibitions from 2 December 2026, and rewrote the AI literacy duty. This tracker lists every application date with its legal basis and current status, so security, privacy and compliance teams can plan from one maintained page. It does not repeat what each obligation requires: for that, see the EU AI Act Compliance Guide. Dates are taken from the consolidated Article 113 and Article 111 published by the European Commission's AI Act Service Desk, the amending regulation on EUR-Lex and the European Parliament procedure file; check those sources before relying on a date.
Last reviewed: 27 October 2026 (consolidated text as at 27 July 2026)
Key takeaways
- The omnibus is law. Regulation (EU) 2026/1744 of 8 July 2026 was published in the Official Journal on 24 July 2026 and entered into force on 27 July 2026. Plans based on the proposal or on the original dates need updating.
- High-risk obligations moved. Chapter III Sections 1 to 3 now apply from 2 December 2027 for Annex III systems (Article 6(2)) and from 2 August 2028 for Annex I systems (Article 6(1)).
- Much already applies. Prohibitions and AI literacy since 2 February 2025, general-purpose AI model obligations and penalties since 2 August 2025, and the general application date, including Article 50 transparency, since 2 August 2026.
- New dates arrive soon. New prohibitions on non-consensual intimate deepfakes and child sexual abuse material, and the end of the Article 50(2) transition for existing generative systems, both fall on 2 December 2026.
- Use the extra time for engineering, not waiting. Logging, human oversight, access control and cybersecurity for high-risk systems take longer to build than to document, and identity controls for AI systems sit at the centre of them.
Status key
- Applies: the provision is in application today (27 October 2026).
- Scheduled: the date is fixed in the consolidated text and has not yet arrived.
- Amended: the date or wording was changed by Regulation (EU) 2026/1744.
- Commission task: a deadline for the Commission or Member States rather than for providers and deployers.
The amending act at a glance
| Step | Date | Source |
|---|---|---|
| Commission proposal COM(2025) 836, Digital Omnibus on AI | 19 November 2025 | European Parliament procedure file 2025/0359(COD) |
| Political agreement between Parliament and Council | 7 May 2026 | European Commission AI policy page |
| Parliament first reading position adopted | 16 June 2026 | Procedure file |
| Council adoption | 29 June 2026 | Procedure file |
| Signature of Regulation (EU) 2026/1744 | 8 July 2026 | Procedure file; EUR-Lex |
| Publication in the Official Journal | 24 July 2026 | Procedure file |
| Entry into force (third day after publication) | 27 July 2026 | Recital 46 of the amending regulation; EUR-Lex metadata |
Dates already in application
| Date | What applies | Legal basis | Status |
|---|---|---|---|
| 1 August 2024 | AI Act (Regulation (EU) 2024/1689) enters into force; no obligations yet | Article 113 | Applies |
| 2 February 2025 | Chapters I and II: definitions, AI literacy (Article 4) and prohibited practices (Article 5) | Article 113(a) | Applies; Article 4 wording amended |
| 2 August 2025 | Notifying authorities and notified bodies (Chapter III Section 4), general-purpose AI models (Chapter V), governance (Chapter VII), penalties (Chapter XII, except Article 101) and confidentiality (Article 78) | Article 113(b) | Applies |
| 27 July 2026 | Articles 102 to 110 (amendments to other Union legislation) | Article 113(d) | Applies; Amended (new point) |
| 2 August 2026 | General application date: everything not listed elsewhere, including Article 50 transparency obligations and Article 101 fines for general-purpose AI model providers | Article 113, second paragraph | Applies |
The AI literacy duty in Article 4 now reads as a requirement for providers and deployers to take measures to support the development of AI literacy, and states that it does not require them to guarantee any specific level of literacy for any individual. See the EU AI Act AI Literacy Obligations (Article 4) Guide.
Scheduled dates for providers and deployers
| Date | What applies | Legal basis | Status |
|---|---|---|---|
| 2 December 2026 | New prohibitions: AI systems that generate or manipulate realistic non-consensual intimate imagery, and systems that generate child sexual abuse material (Article 5(1), points (ba) and (bb), with Article 5(1a) and (1b)) | Article 113(a) as amended | Scheduled; Amended |
| 2 December 2026 | End of the four-month transition for Article 50(2) marking of synthetic content, for providers whose systems were placed on the market before 2 August 2026 | Article 50 as amended; recital 38 | Scheduled; Amended |
| 2 August 2027 | Providers of general-purpose AI models placed on the market before 2 August 2025 must comply | Article 111(3) | Scheduled |
| 2 December 2027 | Chapter III Sections 1 to 3 for high-risk systems under Article 6(2) and Annex III (for example biometrics, critical infrastructure, education, employment, migration and border control) | Article 113(c)(i) as amended | Scheduled; Amended |
| 2 August 2028 | Chapter III Sections 1 to 3 for high-risk systems under Article 6(1) and Annex I (AI in products covered by Union harmonisation legislation, such as lifts and toys) | Article 113(c)(ii) as amended | Scheduled; Amended |
| 2 August 2030 | Providers and deployers of high-risk systems intended to be used by public authorities must have taken the necessary steps to comply | Article 111(2) as amended | Scheduled |
| 31 December 2030 | Components of the large-scale IT systems in Annex X placed on the market or put into service before 2 August 2027 must comply | Article 111(1) | Scheduled |
High-risk systems already on the market before their Chapter III date are covered only if they are subject to significant changes in their design after that date (Article 111(2)); the amending regulation's recital 39 explains that this is assessed for the type and model, using the date the first unit was placed on the market. Public authority systems have the separate 2030 deadline.
Who the dates bind: provider dates apply to whoever places a system on the EU market or puts it into service under its own name, wherever it is established; deployer obligations under Article 26 follow the same Chapter III dates for organisations using high-risk systems in the EU. A deployer that substantially modifies a high-risk system or puts its own name on it can become a provider, which moves it onto the provider obligations for that system, so check role changes when you customise or rebrand a vendor's AI system.
Commission and Member State deadlines
| Date | Task | Legal basis | Status |
|---|---|---|---|
| 2 August 2025 | Member States designate national competent authorities and lay down penalty rules (Chapter VII and XII application) | Article 113(b) | Applies |
| 2 August 2027 | At least one national AI regulatory sandbox operational in each Member State | Article 57(1) as amended | Commission task (Member States) |
| 2 September 2027 | Commission guidance and template for the post-market monitoring plan | Article 72(3) as amended | Commission task |
| 28 January 2028 | Notified bodies notified under Union harmonisation legislation in Annex I Section A apply for designation under the AI Act | Article 43(3) as amended | Scheduled (notified bodies) |
Penalties
| Infringement | Maximum fine | Article |
|---|---|---|
| Prohibited practices (Article 5) | EUR 35 million or 7% of total worldwide annual turnover, whichever is higher | 99(3) |
| Other operator and notified body obligations | EUR 15 million or 3%, whichever is higher | 99(4) |
| Incorrect, incomplete or misleading information to authorities | EUR 7.5 million or 1%, whichever is higher | 99(5) |
| SMEs and start-ups | The lower of the amount or percentage | 99(6) |
| Small mid-caps (new) | The lower of the amount or percentage, for 99(4) and 99(5) only | 99(6a), added by the omnibus |
The omnibus also gives the AI Office enforcement powers with periodic penalty payments of up to 5% of average daily worldwide turnover per day (new Article 75c(5)).
Identity and security obligations on the timeline
The AI Act does not use the term non-human identity, but the high-risk requirements depend on knowing which accounts, services and agents can change, run or query an AI system. When the dates above arrive, these are the requirements identity and security teams will be asked to evidence.
| Requirement | Article | Applies from | Identity and NHI evidence |
|---|---|---|---|
| Record-keeping (automatic logs) | 12 | 2 Dec 2027 (Annex III) / 2 Aug 2028 (Annex I) | Logs that attribute actions to a human user, service account or agent identity; protection of those logs |
| Human oversight | 14 | Same | Who can override or stop the system, with named roles and access controls for that capability |
| Accuracy, robustness and cybersecurity | 15 | Same | Least privilege for model, data and pipeline credentials; protection against unauthorised alteration of training data or the system |
| Deployer obligations (use, monitoring, logs retention) | 26 | Same | Accounts and API keys used to operate the system, owners, and access reviews |
| Transparency for interacting systems and synthetic content | 50 | 2 Aug 2026 (50(2) transition to 2 Dec 2026) | Agents that interact with people must disclose they are AI; protection of marking and detection systems |
| AI literacy measures | 4 | 2 Feb 2025 | Training records for staff who operate AI systems and agents |
For agent-specific identity controls that support these requirements, see the Agentic AI Identity Guide.
Mapping to other frameworks
| AI Act area | ISO/IEC 42001:2023 | NIST AI RMF 1.0 | NIST CSF 2.0 |
|---|---|---|---|
| Risk management and governance roles | AI management system, roles and responsibilities | GOVERN, MAP | GV.RR-02, GV.OC-03 |
| Logging and monitoring (Articles 12, 26, 72) | Operation and monitoring controls | MEASURE, MANAGE | DE.CM-09 |
| Cybersecurity and access (Article 15) | Supported by ISO/IEC 27001:2022 controls | MANAGE | PR.AA-05 |
| AI literacy (Article 4) | Competence and awareness | GOVERN | PR.AT-01 |
Harmonised standards that give a presumption of conformity are still being developed; the conformity route is covered in the EU AI Act Conformity Assessment and Harmonised Standards Guide. See also the NIST AI RMF Guide.
What to watch next
- Article 50 guidelines and code of practice: the Commission published draft Article 50 guidelines dated 7 May 2026, and the omnibus allows an implementing act on detecting and labelling generated content if a code of practice is inadequate. Watch for final versions before 2 December 2026.
- Harmonised standards: the availability of harmonised standards drives how much work conformity assessment will take before the December 2027 and August 2028 dates.
- Post-market monitoring template: due from the Commission by 2 September 2027.
- National sandboxes and authorities: sandboxes due by 2 August 2027; national enforcement practice will shape priorities.
- Further amendments: the omnibus shows the dates can move again. Re-check the consolidated text at each review of this tracker.
Common pitfalls
- Planning to the proposal, not the law. Some commentary still describes the omnibus as pending or quotes proposal wording; the adopted text is Regulation (EU) 2026/1744.
- Treating the delay as a pause. Prohibitions, literacy, general-purpose AI and Article 50 obligations already apply; only Chapter III high-risk dates moved.
- Assuming legacy systems are exempt. A significant design change after the Chapter III date brings an existing high-risk system into scope, and public authority systems have their own 2030 deadline.
- Missing the December 2026 dates. Generative systems already on the market need Article 50(2) marking by 2 December 2026, the same day the new prohibitions apply.
- Leaving logging and access to the end. Articles 12, 14 and 15 need engineering work on identities, logs and oversight that cannot be produced in the final quarter.
How to use this tracker
- Classify each AI system you provide or deploy (prohibited, high-risk under Annex III or Annex I, Article 50 transparency, general-purpose AI model, minimal risk). See the EU AI Act High-Risk AI Systems Guide.
- Confirm your role for each system, provider or deployer, using the EU AI Act Provider vs Deployer Obligations Guide.
- Assign each system its applicable dates from the tables above and record them in your risk register with the legal basis.
- Work backwards from each date: allow time for logging, access control and oversight engineering, testing and evidence collection.
- Re-check the dates against the AI Act Service Desk consolidated text at least quarterly and whenever the Commission publishes guidance.
Practitioner checklist
- Replace any 2 August 2026 high-risk deadline in project plans with 2 December 2027 or 2 August 2028 as applicable.
- Confirm Article 50 disclosure and marking are in place, and that legacy generative systems will meet Article 50(2) by 2 December 2026.
- Check that no system falls under the new Article 5 prohibitions applying from 2 December 2026.
- Record AI literacy measures for staff who operate AI systems, against the amended Article 4 wording.
- Inventory the service accounts, API keys and agent identities each high-risk system uses, with owners.
- Design Article 12 logging so actions are attributable to a named human or non-human identity.
- Track legacy general-purpose AI model compliance by 2 August 2027 if you provide such models.
- Schedule a quarterly review of this tracker against the consolidated text.
Standards and references
- European Commission AI Act Service Desk: Article 113 Entry into Force and Application
- European Commission AI Act Service Desk: Article 111 AI Systems Already Placed on the Market
- European Commission AI Act Service Desk: Timeline for the Implementation of the EU AI Act
- EUR-Lex: Regulation (EU) 2026/1744 (Digital Omnibus on AI)
- EUR-Lex: Regulation (EU) 2024/1689 (Artificial Intelligence Act)
- European Parliament Legislative Observatory: Procedure 2025/0359(COD)
- European Commission: AI Act Regulatory Framework
- European Commission AI Act Service Desk: Article 99 Penalties
This guide summarises the application timeline of the EU Artificial Intelligence Act as amended by Regulation (EU) 2026/1744 and is not legal advice. Related NHI Mgmt Group resources: EU AI Act Compliance Guide · EU AI Act Provider vs Deployer Obligations Guide · EU AI Act Article 50 Transparency Obligations Guide · EU AI Act Prohibited Practices Guide