2026-07-29

EU AI Act 2026 Timeline: How to Study It with Flashcards

On 27 July 2026, the EU AI Act timeline changed six days before its general application date. A timeline card written the previous week could now contain the wrong high-risk date while missing rules that still start on 2 August.

The direct answer is to rebuild the timeline from the amended law, then turn its stable dates and source relationships into a small, versioned deck. The AI Omnibus regulation did not postpone the whole Act. For high-risk systems, it moved the application of Chapter III, Sections 1–3, except Article 6(5), to 2 December 2027 for the Article 6(2)/Annex III pathway and 2 August 2028 for the Article 6(1)/Annex I pathway, subject to Article 2(2)’s limitation for Annex I Section B systems. The Commission’s Article 50 guidelines confirm that those transparency obligations apply from 2 August 2026. A narrow transition gives certain providers of systems already on the market until 2 December 2026 to comply with Article 50(2).

This study workflow uses official EU materials checked on 29 July 2026. Flashcards can help you remember dates, terms, and source relationships. They cannot classify a real system, establish AI literacy or compliance, replace current law and guidance, or provide legal advice. Check the sources again whenever you use the deck for a real decision.

A hand arranges four flashcards beside an open ring binder and a laptop displaying a network diagram.

Put the legal text above every summary

Use a source hierarchy that makes disagreements easy to resolve:

Start with EUR-Lex. Read Regulation (EU) 2026/1744, linked above, together with Regulation (EU) 2024/1689, the AI Act. For a rule changed by the Omnibus, follow the amendment into the relevant article. Use the latest consolidated text when EUR-Lex provides one, and check its consolidation date.
Use the Commission’s AI Act overview to understand the broad structure and implementation timeline.
Open the provision-specific Commission material, such as the Article 50 guidelines, for scope explanations, exceptions, and examples.
Keep your notes and cards at the bottom of the stack. If a summary conflicts with the regulation, the regulation controls.

Put the article or annex, source title or CELEX number, source URL, and retrieval date on every back. Add a version note such as post-2026/1744 when an amendment affects the answer.

AI can shorten a question or flag duplicates. Check every draft against the official source before review, especially dates, exceptions, actor labels, and legal verbs. The cleanup process in How to Fix AI Flashcards applies particularly well here: polished wording deserves no extra trust.

Build one compact EU AI Act timeline first

This timeline gives you the spine of the deck. It does not list every obligation or transition. Each row names the provision or category that gives the date meaning.

Date	What to retrieve
2 February 2025	Chapters I and II began to apply. The Omnibus later replaced Article 4 and assigned the newly added Article 5 prohibitions a separate 2026 start.
2 August 2025	Chapter III, Section 4; Chapters V, VII, and XII; and Article 78 began to apply, except Article 101. This stage includes governance and general-purpose AI model rules.
27 July 2026	Regulation (EU) 2026/1744 entered into force on the third day following its 24 July publication. It revised Article 4, amended the timeline, and added prohibited practices.
2 August 2026	The general application date remains, subject to the exceptions in Article 113. Article 50 transparency obligations apply from this date.
2 December 2026	The newly added Article 5(1), first subparagraph, points (ba) and (bb), and Article 5(1a) and (1b) begin to apply. This is also the Article 50(2) compliance date for providers covered by the new Article 111(4) transition.
2 December 2027	Chapter III, Sections 1–3, except Article 6(5), apply for systems classified as high-risk under Article 6(2) and Annex III.
2 August 2028	Article 113 sets this date for Chapter III, Sections 1–3, except Article 6(5), as regards Article 6(1)/Annex I systems, but Article 2(2) limits AI Act application for Section B products, including machinery.

The dates come from amended Article 113 and transitional Article 111 in Regulation (EU) 2026/1744. The Commission’s 27 July AI Omnibus notice confirms the entry-into-force date and the two high-risk timelines.

Avoid compressing the table into “the AI Act starts in 2026” or “high-risk was delayed.” Either version strips away the structure you need to retrieve.

Turn dates into cards with one unambiguous answer

A date card should ask for one date or one event.

Weak front:

When does the EU AI Act apply?

The Act’s staged application and transitional rules make several answers defensible.

Better cards:

Front: What is the general application date in Article 113 of the EU AI Act after Regulation (EU) 2026/1744?

Back: 2 August 2026, with the exceptions stated in Article 113.

Source: Regulation (EU) 2024/1689, Article 113, as amended by Regulation (EU) 2026/1744, Article 1(40). Retrieved: 29 July 2026. Version: post-Omnibus.

Front: When do Chapter III, Sections 1–3, except Article 6(5), apply to high-risk AI systems classified under Article 6(2) and Annex III?

Back: 2 December 2027.

Source: Regulation (EU) 2026/1744, Article 1(40). Retrieved: 29 July 2026.

Front: Which event occurred on 27 July 2026 in the EU AI Act timeline?

Back: Regulation (EU) 2026/1744, the Digital Omnibus on AI, entered into force.

Source: Regulation (EU) 2026/1744, Article 4 and OJ publication data. Retrieved: 29 July 2026.

Create both directions only when you need both in practice. One is usually enough. How to Make Better Flashcards explains why narrow fronts and short backs produce cleaner review signals.

Learn the risk structure through contrasts

The Commission presents four levels in its public overview: unacceptable risk, high risk, transparency risk, and minimal or no risk. Use that map for orientation, never as a classification shortcut.

Contrast cards work better than isolated definitions because the easy mistake is mixing neighboring structures:

Front: In the Commission’s four-level overview, which level covers specific disclosure duties such as informing people about certain AI interactions or generated content?

Back: Transparency risk. For operative requirements, open Article 50 and the current guidance.

Source: Commission AI Act overview. Retrieved: 29 July 2026.

Front: What is the structural difference between Article 6(1)/Annex I and Article 6(2)/Annex III in the high-risk timeline?

Back: Article 6(1)’s two conditions are a starting point, read with Article 6(1a)–(1c): the AI system is a product or safety component covered by Annex I legislation, and the product requires third-party conformity assessment. Under Article 6(1c), an assessment solely for non-health-and-safety risks does not satisfy the second condition. Article 6(2) covers Annex III systems, subject to Article 6(3). Article 113 sets the pathway dates—2 August 2028 and 2 December 2027—but Article 2(2) limits provisions applying to Section B product systems.

Source: Regulation (EU) 2024/1689, Articles 2, 6, and 113, as amended by Regulation (EU) 2026/1744, Article 1(2), (8), (40), and (41). Retrieved: 29 July 2026.

That second card teaches navigation. It does not classify a product. A real assessment also needs the system’s intended purpose, technical design, deployment context, applicable product law, and current guidance.

Keep actors, objects, and actions on separate cards

Legal sentences become much easier to remember when you stop storing every noun in one prompt.

Start with the exact Article 3 actors:

a provider develops an AI system or general-purpose AI model, or has one developed, and places it on the market or puts the AI system into service under its own name or trademark, whether for payment or free of charge
a deployer uses an AI system under its authority, outside the personal non-professional exception in the definition

Then separate the objects. An AI system and a general-purpose AI model are different legal terms, and different rules attach to each.

Finally, learn the actions independently. “Placing on the market” concerns first making an AI system or GPAI model available on the Union market. “Putting into service” concerns supplying an AI system for first use directly to a deployer or for own use in the Union for its intended purpose. “Use” is a different action again.

A clean card pair looks like this:

Front: Under Article 3, which actor uses an AI system under its authority, excluding personal non-professional use?

Back: The deployer. Source: Regulation (EU) 2024/1689, Article 3(4). Retrieved: 29 July 2026.

Front: Does an AI Act guideline have the same legal role as an obligation in the regulation?

Back: No. The regulation contains the obligation. Commission guidelines offer practical guidance on its application. Check the guideline’s status, scope, and date, and return to the regulation for the legal text. Retrieved: 29 July 2026.

Separate obligation, guidance, code of practice, and informational resource too. The document type tells you what weight to give the answer.

Make Article 50 cards point back to the source

Article 50 shows why scenario cards need restraint. The Commission’s Article 50 guidelines linked above say the transparency obligations apply from 2 August 2026. Its Article 50 transparency overview separates provider duties from deployer disclosures and points readers to the exceptions.

Build cards that send you to the relevant paragraph:

Front: A provider is checking machine-readable marking for outputs from a system generating synthetic audio, images, video, or text. Which provision should be opened first?

Back: Article 50(2), including its standard-editing and law-enforcement exceptions, followed by the current guidelines. This card does not determine whether the system or output is in scope. Retrieved: 29 July 2026.

Front: A deployer is checking disclosure for image, audio, or video content that may constitute a deepfake. Which provision should be opened first?

Back: Article 50(4), including the law-enforcement exception and the rule for evidently artistic, creative, satirical, fictional, or analogous works. No scope conclusion is stored on this card. Retrieved: 29 July 2026.

Front: What is the narrow 2 December 2026 transition added for Article 50?

Back: Providers of AI systems, including GPAI systems, generating synthetic audio, image, video, or text content that were placed on the market before 2 August 2026 must take the necessary steps to comply with Article 50(2) by 2 December 2026.

Source: Regulation (EU) 2026/1744, amendment to Article 111. Retrieved: 29 July 2026.

Do not rewrite that transition as “Article 50 was delayed until December.” The Commission guidance and overview both state that Article 50 applies from 2 August 2026. The December provision is limited by actor, obligation, system function, and pre-2-August market timing.

Version the Omnibus changes instead of stacking old cards

The AI Omnibus 2026 made an undated deck risky.

Use a changed tag only for a precise amendment:

relevant Chapter III high-risk dates split between Annex III and Annex I
Article 4 now requires providers and deployers to take measures supporting the development of AI literacy, while expressly saying they need not guarantee a specific level for an individual
specified new prohibitions apply from 2 December 2026
a narrow Article 50(2) transition was added for qualifying systems placed on the market before 2 August 2026

Use an unchanged tag where an old summary is likely to overreact:

2 August 2026 remains the general application date, with exceptions
Article 50 transparency obligations apply from 2 August 2026

The amended AI literacy wording appears in Regulation (EU) 2026/1744. A useful card asks what Article 4 now requires: measures supporting AI literacy among relevant staff and other persons dealing with the operation and use of AI systems on a provider’s or deployer’s behalf, without a guaranteed level for any individual. Recall of that wording does not show that an organization’s measures satisfy Article 4.

Date every amendment-sensitive card. When a source changes, edit or delete the old card instead of keeping contradictory old and new answers.

Put the verified deck into an FSRS workflow

Create one deck named EU AI Act — official sources. Use tags to create useful slices without duplicating cards:

timeline, actor, risk, article-50, ai-literacy, omnibus, needs-source-check

Keep each item as a plain front/back card. Put the short answer first on the back, followed by the article or annex, official source, retrieval date, and version note. Anything tagged needs-source-check stays outside normal review until you verify it. Treat that tag as quarantine.

Flashcards Open Source App features include plain front/back cards, decks and tags, and FSRS review. The hosted app is free, and the project also has an open-source self-hosted path. Once a card passes its source check, let FSRS-6 schedule it. Attempt the answer before revealing the back, then choose Again, Hard, Good, or Easy based on the recall you actually produced:

Again for a missing or wrong required answer
Hard for correct recall with serious difficulty
Good for correct recall with ordinary effort
Easy for effortless correct recall

Active Recall vs Spaced Repetition explains the division: retrieval exposes what you can produce now; spacing decides when the verified card returns. The getting-started guide covers the hosted and self-hosted routes if you want to put the workflow into practice. Flashcards does not supply an official EU AI Act deck, verify legal content, or update your cards when the law changes.

Keep legal judgment outside the review queue

Flashcards work well when the answer stays small and source-bound. A factual investigation or a chain of legal interpretation belongs outside the review queue.

Do not use a card to decide:

whether a particular tool meets the definition of an AI system
whether your organization is a provider, deployer, importer, distributor, or several operators in different contexts
whether a system falls under Article 6(1), Article 6(2), an Annex III qualification, or another rule
whether an Article 50 exception applies to a specific output or publication
whether training, documentation, controls, disclosures, or other measures establish compliance

Those questions require current law, guidance, documented facts, and often qualified legal advice. The answer can change with the product, intended purpose, distribution model, or use context.

Use scenario cards to practice finding the relevant provision. Reach conclusions from the current sources, documented facts, and appropriate advice. The same split appears in Flashcards vs Practice Tests: cards keep components retrievable, while larger exercises test whether you can use them together. What Should Go on a Flashcard? helps cut anything too broad for honest grading.

FAQ about the EU AI Act 2026 timeline

Was the whole EU AI Act postponed?

No. Regulation (EU) 2026/1744 made targeted amendments. The general application date remains 2 August 2026, with exceptions. The relevant Chapter III high-risk provisions now apply on later dates for the two specified pathways.

When does Article 50 of the AI Act apply?

Article 50 applies from 2 August 2026. The 2 December 2026 date is a narrow transition for Article 50(2) providers of qualifying synthetic-content-generating systems placed on the market before 2 August 2026. It is not a general Article 50 postponement.

What changed in the AI Act high-risk timeline?

Article 113 sets 2 December 2027 for Article 6(2)/Annex III high-risk systems and 2 August 2028 for Article 6(1)/Annex I high-risk systems. Keep the pathway and Article 2(2) limitation on the same card.

Does completing an EU AI Act flashcard deck prove compliance or AI literacy?

No. A deck records study activity and recall. It does not establish legal scope, prove that measures meet Article 4, document operational controls, or demonstrate compliance with other obligations.

How often should I update the deck?

Check official sources before study tied to a decision, assessment, training update, or deadline. For ordinary learning, check monthly and whenever EUR-Lex or the Commission publishes an amendment, guideline, code, or correction. Record the retrieval date each time.

Can AI generate the cards?

AI can draft fronts, shorten backs, and find duplicates. Verify every legal claim against EUR-Lex or the relevant official Commission material before review. Open every cited source yourself; an AI-generated citation is not evidence.

Start with five verified EU AI Act timeline cards: 27 July 2026, 2 August 2026, 2 December 2026, 2 December 2027, and 2 August 2028. Open the current EUR-Lex text, then add the provision, official source, retrieval date, and version to every back. Once those five are accurate, expand into actors, risk pathways, Article 50, and AI literacy. Keep every answer small enough to grade and every legal conclusion outside the deck.