Short answers to the questions people ask most about Article 50 of the EU AI Act. Each one links to a full explainer with the detail, the citations and the edge cases.
Usually — if an AI chatbot interacts directly with visitors, Article 50(1) requires telling them it's an AI, unless that's obvious from context.
Under two different rules: machine-readable marking by the AI provider (50(2)), and a visible disclosure by the publisher for public-interest text that hasn't had human editorial review (50(4)).
Often under more than one rule — 50(1) if it talks with people, 50(2) marking by the tool that generates it, and 50(4) if it's realistic enough to pass as a real person.
It can: Article 2 reaches non-EU providers and deployers whose AI system's output is used in the EU, though how that applies to websites merely accessible from the EU isn't settled.
Yes — Article 50 cares that a disclosure is present, clear and accurate, not who or what drafted it.
Fines of up to €15 million or 3% of worldwide annual turnover under Article 99(4), with a lower cap for SMEs and small mid-cap enterprises.
The AI Liability Directive was withdrawn, so claims run through the revised Product Liability Directive, national tort law, and AI Act non-compliance as evidence.
Provider, deployer, deep fake, machine-readable marking — the Article 50 glossary defines each one, with citations.
Article50.io is an automated Article 50 transparency assessment platform that scans websites for potential EU AI Act transparency obligations and provides remediation guidance, implementation instructions, and compliance-ready disclosure language.
The free scan shows your single most severe finding in about 30 seconds — no signup, public pages only.
Automated technical guidance, not legal advice. Citations refer to Regulation (EU) 2024/1689.