§ Regulation (EU) 2024/1689 — Article 50
The EU AI Act transparency deadline is August 2, 2026.
From that date, AI-generated content published in the EU must be disclosed — deepfakes labelled, chatbots announced, generative outputs marked machine-readably. Fines reach €15M or 3% of worldwide turnover.
§ 01 — Who must do what
Four obligations, one date
Anyone deploying chatbots or AI systems that interact with people
Users must be told they are interacting with an AI system, unless it's obvious from context.
Providers of generative AI (text, image, audio, video)
Outputs must be marked as artificially generated in a machine-readable format (e.g. provenance metadata, watermarking).
Anyone publishing deepfakes — AI-generated or AI-manipulated realistic media
Must clearly and distinguishably disclose that the content has been artificially generated or manipulated.
Anyone publishing AI-generated text to inform the public on matters of public interest
Must disclose the text was artificially generated, unless it underwent human editorial review with responsibility.
Source: EU AI Act, Article 50 (transparency obligations). Existing generative AI systems have until 2 Dec 2026 for machine-readable marking, per the May 2026 AI Omnibus.
§ 02 — Readiness checklist
Get compliant before the deadline
Inventory your AI content
List every place you publish AI-generated or AI-edited media: site, YouTube, TikTok, Instagram, X, LinkedIn, newsletters, ads.
Classify how much AI is involved
Fully AI-generated, AI-edited real media, or AI-assisted human-led — each triggers different obligations.
Get the exact label for each case
Our free generator tells you whether disclosure is required, under which rule, and hands you ready-to-paste wording.
Make it machine-readable where required
For providers, Article 50(2) requires marking outputs in a machine-readable format — provenance metadata, not just a caption.
Seal it and keep evidence
Attach an Authenticity Badge with a public verify page, so your disclosure is documented and checkable.
§ 03 — Questions, answered
EU AI Act deadline — FAQ
When does the EU AI Act require AI content to be labelled?
The transparency obligations in Article 50 of the EU AI Act apply from 2 August 2026. They cover chatbot disclosure, machine-readable marking of generative AI outputs, deepfake disclosure, and disclosure of AI-generated text published on matters of public interest. Existing generative AI systems have until 2 December 2026 to add machine-readable marking, per the May 2026 AI Omnibus.
Who has to comply — companies or individual creators?
Both can be in scope. Providers of generative AI systems must mark outputs in a machine-readable way; deployers — which can include businesses and individual creators publishing deepfakes or public-interest AI text in the EU — must disclose clearly. Purely personal, non-professional use benefits from lighter treatment, but published realistic synthetic media is squarely covered.
What happens if I don't disclose?
Non-compliance with Article 50 transparency obligations can be fined up to €15 million or 3% of total worldwide annual turnover, whichever is higher (Article 99(4) EU AI Act). National market-surveillance authorities enforce it.
Does this apply to content made before August 2026?
The obligations attach to systems and content made available from the application date; existing generative AI systems get a transition until 2 December 2026 for machine-readable marking. If older AI content is still published and qualifies as a deepfake, disclosing it is the safe course.
Is a hashtag like #AI enough?
For deployers, the disclosure must be clear and distinguishable — a buried hashtag is risky. For providers, Article 50(2) requires machine-readable marking, which a hashtag is not. Use explicit wording plus provenance metadata where applicable.
Disclosa provides informational tooling, not legal advice. Rules change frequently and vary by context. Always verify against the official source and consult a qualified professional for your case.