A quick reference for the EU AI Act after the Digital Omnibus. Save this page as a PDF for your team, or use it as a training handout.
| Date | What happens | Status |
|---|---|---|
| 1 Aug 2024 | AI Act enters into force | Done |
| 2 Feb 2025 | Prohibited practices and AI literacy apply | In force |
| 2 Aug 2025 | GPAI model rules and governance apply | In force |
| 27 Jul 2026 | Digital Omnibus (Regulation (EU) 2026/1744) enters into force | In force |
| 2 Aug 2026 | Transparency duties apply (Article 50) | In force |
| 2 Dec 2026 | New prohibitions apply (intimate imagery, CSAM) | New |
| 2 Dec 2027 | High-risk Annex III duties apply, moved from Aug 2026 | Deferred |
| 2 Aug 2028 | High-risk Annex I duties apply, moved from Aug 2027 | Deferred |
| Breach | Maximum fine |
|---|---|
| Prohibited practices (Article 5) | Up to €35 million or 7 percent of total worldwide annual turnover, whichever is higher |
| Most other obligations, including high-risk and transparency | Up to €15 million or 3 percent |
| Supplying incorrect, incomplete, or misleading information to authorities | Up to €7.5 million or 1 percent |
For SMEs and start-ups, the lower of the fixed amount or the percentage applies.
| Provider | An organization that develops an AI system or general-purpose model and places it on the market under its own name. |
| Deployer | An organization that uses an AI system under its own authority in a professional setting. Most organizations are deployers. |
| GPAI | A general-purpose AI model that can handle many tasks and be built into other systems. Its providers carry documentation and transparency duties. |
| Annex III | The list of high-risk use areas, including employment, credit, biometrics, education, and essential services. |
| Conformity assessment | The check that a high-risk system meets the Act's requirements before it goes on the market. |
| FRIA | A fundamental rights impact assessment, required of certain deployers of high-risk AI (Article 27). |
| DPIA | A data protection impact assessment under the GDPR, often relevant alongside AI use. |
| Systemic risk | A category of the most capable general-purpose models, which carry extra obligations. |
| AI literacy | A suitable level of understanding of AI among the people who use it, which providers and deployers must support (Article 4). |
| Transparency obligation | The duty to disclose AI interaction and label AI-generated or manipulated content (Article 50). |
| Digital Omnibus | The 2026 simplification package that deferred the high-risk deadlines and adjusted several other duties. Published as Regulation (EU) 2026/1744 in the Official Journal on 24 July 2026, in force since 27 July 2026. |
The AI Act transparency duty (Article 50) has applied since 2 August 2026. Where you publish AI-generated or manipulated content that a person could take as authentic, you have to disclose it. The European Commission has released an official set of icons that make that disclosure clear and consistent. They are free to use, come in three types and four colour variations, and were user-tested, with results improving when the icon is paired with a short text label such as "modified".
| Icon | When to use it |
|---|---|
| Basic | When AI was involved in creating deepfake content (image, audio, or video) or published text, often paired with a text label such as "voices generated with". |
| Fully AI-Generated | When the whole item is generated by AI with no human-created elements or editorial control beyond prompting, for example a fully AI-made video or an AI-written news summary. |
| Partially AI-Modified | When human-made content is partially changed with AI, for example a face swapped into an authentic photo, or an empty room furnished with AI. |
Official files hosted by the European Commission. © European Union, 2026. Source: EU icons for labelling AI-generated content. The icons are optional, and using them does not by itself establish compliance with Article 50.
Does AI-assisted writing need this label? Generally no. Article 50(4) exempts text that has undergone human review or editorial control, where a person or organization takes editorial responsibility for it. A draft that a person reads, edits, and sends is not the published, unreviewed, public-interest content the labelling duty targets. These icons are for content presented to the public as AI-generated, not for everyday assisted writing.
As of August 2026. This reference is for orientation and internal training. It is not legal advice and does not certify compliance, and it simplifies the underlying text. The Digital Omnibus was published in the Official Journal on 24 July 2026 as Regulation (EU) 2026/1744 and entered into force on 27 July 2026. Confirm specifics with qualified counsel.