The EU AI Act’s transparency rules, in plain language.
We write these guides for business owners, not lawyers. Every claim cites the official text — Regulation (EU) 2024/1689 — so you can verify us — alongside the Commission’s official interpretation: the Article 50 Guidelines (20 July 2026) and the Code of Practice on Transparency of AI-generated Content (10 June 2026). Where something genuinely remains unsettled — mainly the evolving technical marking standards — we say “this is not yet settled” instead of guessing.
EU AI Act Article 50, explained in plain language
The whole transparency article, duty by duty — the pillar guide.
Chatbot disclosure: what your chatbot must say
Art. 50(1), the channels people forget, and the “obvious” exemption.
Synthetic content marking: the rules now in force
Art. 50(2): whose job the machine-readable marks are, and what to do now.
Does the AI Act apply to my small business?
Honestly: partly. What realistically hits an SME — and what doesn’t.
AI-generated images: labels and the law
Machine-readable marks vs visible labels, and the grey zones, flagged.
“Deepfake” doesn’t mean malicious
Art. 50(4): when AI content showing real things needs a visible label.
AI-written articles and the “public interest” rule
Including the human-editorial exemption most guides skip.
The “obvious AI” exemption, examined
Why the bar is higher than you think — and the economics of relying on it.
AI Act fines: the real numbers
Art. 99 without the fear-mongering: ceilings, SME proportionality, reality.
WhatsApp and phone AI assistants
The duty follows the AI, not the website. How to disclose per channel.