EU AI Act Article 50: chatbot and AI-content disclosure checklist (from 2 Aug 2026)
Short answer: from 2 August 2026, Article 50 of the EU AI Act makes deployers and providers tell people when they are dealing with an AI system or with certain AI-generated content. If you run a customer chatbot, a voice bot, or publish AI images, audio or video that could mislead, you need a clear notice and a record of which systems trigger it.
These duties already apply. AI literacy for staff (Article 4) since 2 Feb 2025; chatbot and AI-content transparency (Article 50) since 2 Aug 2026 (Regulation (EU) 2024/1689).
Free first step: the risk-tier checker and the EURO AI Register run in your browser, free for your first AI system. Nothing is uploaded.
Honest comparison: published SME compliance software starts at €499 a month (CompliPath). The EU AI Act Readiness Kit is US$29 once: an Excel AI inventory and risk-screening workbook, an AI literacy policy template and a plain-English guide. It is the first organising step, not a substitute for a lawyer.
No risk: 14-day money-back if it isn't useful. Not legal advice.
What Article 50 covers
| Situation | Who acts | What people must be told |
|---|---|---|
| People interact with an AI system (chatbot, voice assistant, support bot) | Deployer (you, if you put it in front of customers or staff) | That they are interacting with an AI system, unless it is obvious from the context |
| AI-generated or manipulated images, audio or video that resembles real people, places or events (deepfake-style) | Deployer who publishes or uses it | That the content is artificially generated or manipulated, in a way that is clear and distinguishable |
| AI-generated text published to inform the public on matters of public interest | Deployer, with limited exceptions for editorial control under human review | That the text was generated or manipulated by AI |
| Emotion recognition or biometric categorisation in publicly accessible spaces | Deployer | People in those spaces must be informed |
Read the official text before you rely on a paraphrase: Regulation (EU) 2024/1689, Article 50, on EUR-Lex. High-risk duties for many Annex III uses were postponed by the Digital Omnibus (Regulation (EU) 2026/1744); the transparency clock for Article 50 still runs from 2 August 2026 unless your counsel says otherwise for your case.
A practical checklist
- List every customer-facing or staff-facing chatbot, voice bot and AI writing or media tool.
- Mark which ones talk to people without a human in the loop for the first reply.
- Write the notice wording (for example: "You are chatting with an automated assistant. Ask for a human at any time.") and where it appears: first message, footer, voice prompt.
- For marketing or social media, mark any AI images, audio or video that could be mistaken for a real recording, and add a visible label before publish.
- Keep a dated note of who approved the wording and which systems it covers.
- Re-check the list when you add a new tool or change a vendor model.
Where to keep the record
A spreadsheet is enough. If you want the transparency flags, duties and dates next to each system, the EURO AI Register free edition runs in the browser with no account and no upload (1 AI system plus the Art. 4 AI-literacy record). It can generate a short AI use policy from your entries. A paid membership removes the limits; you do not need it to follow this checklist.
This guide is general information, not legal advice. A checklist or register is a record-keeping aid, not proof of compliance. Unadhi is not affiliated with the EU or any EU authority.
Related: what to put in an AI inventory · what of the Act applies now
Unadhi Guides are free, independent explainers. We correct errors as soon as we find them: write to mininglife7@gmail.com.