8/25/2026, 1:05:02 PM · evaluation-safety

EU AI Act Article 50 Takes Effect: Synthetic-Content Disclosure Rules Now Enforceable

The European Union's first major AI transparency enforcement wave arrived on August 2, 2026, compelling providers and deployers of AI systems to mark synthetic outputs and disclose deepfakes to end users.

Background

<cite index="3-2,3-3">The EU (European Union) AI Act (Regulation (EU) 2024/1689) established the first comprehensive regulatory framework governing the development, deployment, and use of artificial intelligence systems. It entered into force on August 1, 2024, with a phased implementation timeline extending through August 2027.</cite>

<cite index="6-1,6-2">Article 50 of the AI Act applies from August 2, 2026. It sets out transparency obligations for providers and deployers of certain AI systems, including generative and interactive AI systems and deepfakes.</cite>

What Article 50 Requires

The provision establishes a layered duty split between system providers and deployers.

<cite index="14-1">Providers of AI systems — including general-purpose AI systems — that generate synthetic audio, image, video, or text must ensure that outputs are both marked in a machine-readable format and detectable as AI-generated.</cite> <cite index="16-4">The regulation does not specify a single technical mechanism, but points toward standards developed by recognized bodies, which in practice means C2PA and the draft ISO 22144.</cite>

<cite index="2-10">Deployers using AI to create deepfakes — defined in Article 3(60) as AI-generated or manipulated image, audio, or video content that resembles existing persons, objects, places, entities, or events and would falsely appear authentic or truthful — must disclose this.</cite> <cite index="8-10">All disclosures must land no later than the first interaction or exposure, in a manner that is plain, distinguishable, and accessible under existing accessibility rules.</cite>

<cite index="14-10">Where deepfake content forms part of an evidently artistic, creative, satirical, fictional, or analogous work or programme, the disclosure obligation is reduced: it is limited to disclosing the existence of the generated or manipulated content "in an appropriate manner that does not hamper the display or enjoyment of the work."</cite>

<cite index="14-7">Deployers publishing AI-generated text with the purpose of informing the public on matters of public interest must disclose that the text is AI-generated, unless it has been subject to human review and editorial responsibility.</cite>

<cite index="3-7">Deployers must not remove or alter AI watermarks embedded by providers.</cite>

Relationship to the Digital Omnibus

<cite index="1-2">Unlike the Annex III high-risk compliance timeline, which the Digital Omnibus package pushed back to December 2, 2027, Article 50 was left out of that deferral: its core transparency and disclosure duties applied on schedule from August 2, 2026, and national market surveillance authorities can enforce them from that date.</cite>

<cite index="1-3">A four-month grace period was granted specifically for watermarking implementation, running through December 2, 2026.</cite> <cite index="1-4">Organizations relying on that grace period should confirm which specific sub-obligation it covers, since Article 50's disclosure duties for interactive systems and deployer-side deepfake disclosure took effect immediately on August 2, 2026, regardless of the separate watermarking implementation runway.</cite>

Enforcement and Penalties

<cite index="21-6,21-7">August 2, 2026 is not merely when the transparency rules apply — it is when the regulator gains the standing to fine for violating them. The substantive obligation and the enforcement power arrive together.</cite>

<cite index="22-1,22-2">A breach can lead to a fine of up to €15 million or 3% of total worldwide annual turnover, whichever is higher. Enforcement sits with the national market surveillance authorities.</cite>

<cite index="20-2">Compliance with the rules will mainly be enforced by national competent market surveillance authorities.</cite> <cite index="28-2">The AI Act's extraterritorial reach — applying to any provider or deployer whose AI output is used in the EU — creates a powerful "Brussels Effect."</cite>

<cite index="4-5">Content generated prior to August 2, 2026, does not need to be labelled retroactively.</cite>

Supporting Instruments

<cite index="7-1,7-2">A Code of Practice supports compliance with the AI Act transparency obligations related to marking and labelling of AI-generated content. The obligations under Article 50 address risks of deception and manipulation, fostering the integrity of the information ecosystem.</cite> <cite index="17-8,17-9">Even though adherence to the code is voluntary, the transparency requirements under Article 50 are legal obligations. The Commission and the AI Board have confirmed that the code is an adequate voluntary tool to demonstrate compliance.</cite>

<cite index="3-4">All transparency, disclosure, watermarking, and labelling obligations are enforceable, with Commission Guidelines finalized July 20, 2026, and together with the Code of Practice, governing the disclosure requirements.</cite>

Sources

  1. [1]
    EU AI Act Article 50: Transparency Obligations Take Effect – Lab Space
  2. [2]
    The EU AI Act’s Transparency Rules: A Practical Guide to Article 50 | EU Artificial Intelligence Act
  3. [3]
    EU AI Act Article 50 - Shibolet & Co. Law Firm
  4. [4]
    Transparency obligations under Article 50 of the AI Act | Shaping Europe’s digital future
  5. [5]
    European Commission Publishes Draft Code of Practice on AI Labelling and Transparency | Insights | Jones Day
  6. [6]
    Guidelines on transparency obligations for providers and deployers of certain AI systems | Shaping Europe’s digital future
  7. [7]
    Code of Practice on Transparency of AI-generated Content | Shaping Europe’s digital future
  8. [8]
    EU AI Act Article 50 transparency rules enter force
  9. [9]
    arxiv.org
  10. [10]
    EU AI Act Article 50 deepfake rules | Resemble AI
  11. [11]
    What the EU’s New AI Code of Practice Means for Labeling Deepfakes | TechPolicy.Press
  12. [12]
    Article 50: Transparency Obligations for Providers and Deployers of Certain AI Systems | EU Artificial Intelligence Act
  13. [13]
    What constitutes a Deep Fake? The blurry line between legitimate processing and manipulation under the EU AI Act
  14. [14]
    Taking the EU AI Act to Practice Understanding the Draft Transparency Code of Practice - Bird & Bird
  15. [15]
    Verifiable Provenance and Watermarking for Generative AI: An Evidentiary Framework for International Operational Law and Domestic Courts
  16. [16]
    Seeing Is No Longer Believing: Frontier Image Generation Models, Synthetic Visual Evidence, and Real-World Risk
  17. [17]
    When AI Eats Itself: On the Caveats of AI Autophagy
  18. [18]
    What Actually Comes Due on August 2, 2026: EU AI Act Article 50 Transparency and the Digital Omnibus Reset | ComplianceHub.Wiki
  19. [19]
    What if you do not comply with Article 50: enforcement and fines from 2 August 2026
  20. [20]
    What if you do not comply with Article 50: enforcement and fines since 2 August 2026
  21. [21]
    Article 99: Penalties | EU Artificial Intelligence Act
  22. [22]
    Legalithm
  23. [23]
    Implementation Timeline | EU Artificial Intelligence Act
  24. [24]
    Enforcement / fines in the European Union - AI Laws of the World
  25. [25]
    EU AI Act Enforcement: August 2026 Compliance Deadline Explained | ai | informed, clearly