On 2 August 2026, the EU AI Act reached its broadest application stage so far. Article 50 transparency duties began applying, measures supporting innovation became operational, and enforcement powers started for provisions that were already applicable, including prohibited practices, AI literacy, general-purpose AI model rules and transparency duties. The date did not make every high-risk requirement immediately applicable: later milestones still govern new December 2026 prohibitions, Annex III high-risk systems and Annex I product-related high-risk systems, under the timeline as amended by Regulation (EU) 2026/1744.
Why 2 August 2026 Was a Major AI Act Milestone
The AI Act entered into force in 2024, but its obligations have applied in stages. By 2 August 2026, the earlier February 2025 and August 2025 milestones were already in effect, covering definitions, AI literacy, prohibited practices, general-purpose AI model obligations and governance. The August 2026 date added a much broader operational layer. Organisations could no longer treat the Act mainly as a future planning exercise. They needed to distinguish requirements that were already enforceable from high-risk rules that still had later application dates, as amended by Regulation (EU) 2026/1744, and maintain evidence showing how each system, model and operator role had been assessed.
Article 50 Transparency Requirements Began Applying
Article 50 transparency requirements began applying on 2 August 2026. These rules are targeted rather than universal. Depending on the system and use, they can concern informing people that they are interacting with AI, technical marking of synthetic outputs, and disclosures by deployers in specified uses such as deepfakes or certain public-interest text. Providers and deployers should not rely on a single generic AI notice. The correct obligation depends on who provides the system, who uses the output, what type of content or interaction is involved and whether an exception or transition applies.
For the broader regulatory framework, see EU AI Act timeline.
AI Act Enforcement Also Began
The same date also marked the start of important enforcement powers at national and EU level for provisions already in application. This matters because an obligation can exist before a particular enforcement mechanism becomes available, while other obligations remain outside enforcement until their own application date. From 2 August 2026, authorities could enforce rules concerning prohibited practices, transparency, AI literacy and general-purpose AI models within their respective competences. Compliance teams should therefore connect each obligation to its application date, responsible operator, evidence trail and relevant supervisory authority rather than using one blanket 'AI Act effective date'.
The Article 50(2) Transition Did Not End Yet
The August 2026 milestone did not eliminate every transition. The official implementation timeline gives certain providers of AI systems, including some systems based on general-purpose AI models, that were already placed on the market before 2 August 2026 until 2 December 2026 to comply with the Article 50(2) marking and detection obligation for synthetic content. That transition is narrow. It should not be read as postponing all Article 50 duties or all AI Act requirements for existing systems. Teams should identify the exact system version, market date, output type and provider role before relying on the transition.
Who Enforces the AI Act From This Stage?
Enforcement is divided between the AI Office and authorities in the Member States, with the allocation depending on the type of system or model and the provision involved. The AI Office has important responsibilities for general-purpose AI models and specified related systems, while national competent authorities play a central role in supervision and enforcement across the wider Act. Businesses do not need to predict every jurisdictional question before acting, but they should know which legal entity is the provider or deployer, where the system is offered or used, and which authority may be relevant if a compliance issue arises.
Continue with what changes on 2 December 2026.
Which AI Act Deadlines Still Came Later?
Several significant milestones remained in the future after 2 August 2026. New prohibitions concerning specified harmful synthetic-content uses and the Article 50(2) transition arrive on 2 December 2026, under Article 5(1)(ba) and Article 5(1)(bb) added by Regulation (EU) 2026/1744. Member States should have at least one AI regulatory sandbox operational by 2 August 2027, in the Service Desk wording. Rules for high-risk AI systems listed in Annex III apply from 2 December 2027. Rules for high-risk AI embedded in regulated products covered by Annex I apply from 2 August 2028. These later dates are important, but they do not suspend obligations that are already applicable. A system can therefore have a future high-risk deadline while still being subject today to literacy, prohibition, GPAI or transparency rules.
What Organisations Should Do After 2 August 2026
A practical 2026 review should begin with an inventory of AI systems and general-purpose models, then record intended purpose, provider and deployer roles, deployment geography and affected people. Screen each use for prohibited practices and Article 50 transparency requirements, identify any general-purpose AI model obligations, and record which rules are already enforceable. Separately flag systems that may become high-risk under Annex III or Annex I so later implementation work has an owner and deadline. Keep the source provision, decision rationale, reviewer and review date with the record so changes in guidance, system purpose or functionality can trigger reassessment.
Frequently asked questions
Did the entire AI Act become enforceable on 2 August 2026?
No. Enforcement began for important provisions already in application, but later dates still apply to new December 2026 prohibitions and to the main Annex III and Annex I high-risk system rules, as amended by Regulation (EU) 2026/1744.
Did Article 50 start applying on 2 August 2026?
Yes. Article 50 transparency requirements began applying on that date, subject to the specific conditions, exceptions and the limited transition for certain systems already on the market, which ends on 2 December 2026.
Are Annex III high-risk systems already fully subject to the high-risk rules in September 2026?
No. Under the current amended timeline, the Annex III high-risk rules apply from 2 December 2027, as amended by Regulation (EU) 2026/1744.
What does this article not decide?
This article does not decide whether any particular system is subject to a specific Article 50 duty or enforcement action. Those questions turn on the system's function, the operator role and the exact legal wording, and should be confirmed with a competent authority or legal adviser where a decision depends on them.
Continue through this AI Act cluster
This article is an informational overview, not a legal determination. Check the current legal text and official implementation guidance for your system and role.
Consolidated Regulation (EU) 2024/1689, as amended ↗AI Act Service Desk implementation timeline ↗AI Act Service Desk: Article 50 ↗