Article 50 — Transparency obligations for providers and deployers of certain AI systems
AI Act, Regulation (EU) 2024/1689
What does AI Act Article 50 say?
Short answer, straight from the official text
1. Providers shall ensure that AI systems intended to interact directly with natural persons are designed and developed in such a way that the natural persons concerned are informed that they are interacting with an AI system, unless this is obvious from the point of view of a natural person who is reasonably well-informed …
Source: AI Act, Regulation (EU) 2024/1689, read 2026-08-14.
| In the act | EU 2024/1689 |
| Chapter | IV · Transparency obligations for providers and deployers of certain AI systems |
| CELEX | 32024R1689 |
| Read on | 2026-08-14 |
Official text
Read from the EU Publications Office for this CELEX number. The wording stands as published; nothing here is rewritten or summarised.
1. Providers shall ensure that AI systems intended to interact directly with natural persons are designed and developed in such a way that the natural persons concerned are informed that they are interacting with an AI system, unless this is obvious from the point of view of a natural person who is reasonably well-informed, observant and circumspect, taking into account the circumstances and the context of use. This obligation shall not apply to AI systems authorised by law to detect, prevent, investigate or prosecute criminal offences, subject to appropriate safeguards for the rights and freedoms of third parties, unless those systems are available for the public to report a criminal offence.
The article continues in the official text.
Decision guide
Who is affected?
- Providers of AI systems that interact directly with people, such as chatbots. (Art. 50(1))
- Providers of AI systems that generate synthetic audio, images, video or text. (Art. 50(2))
- Deployers of emotion recognition, biometric categorisation and deepfakes, and deployers publishing text created with AI on matters of public interest. (Art. 50(3)–(4))
Questions and answers
Must a chatbot say that it is AI? Yes. People must be told they are interacting with an AI system, unless this is obvious to a reasonably well-informed person from the circumstances. (Art. 50(1))
Must content created with AI be marked? Yes. Providers must ensure outputs are marked in a machine-readable format and detectable as artificially generated or manipulated. (Art. 50(2))
Must deepfakes be disclosed? Yes. Deployers must disclose that the content is artificially generated or manipulated. For evidently artistic, satirical or fictional work the duty is limited to disclosure that does not hamper the work. (Art. 50(4))
When must the information be given? Clearly and distinguishably, at the latest at the first interaction or exposure. (Art. 50(5))
Duties
| Who | What | From |
|---|---|---|
| Provider | Design the system so people are informed they are interacting with AI. (Art. 50(1)) | 2 August 2026 |
| Provider | Mark generated output machine-readably as artificially generated. (Art. 50(2)) | 2 August 2026; 2 December 2026 for systems placed on the market before 2 August 2026 (Art. 111(4)) |
| Deployer | Disclose deepfakes and text created with AI published to inform the public, unless it has undergone human review or editorial control and someone holds editorial responsibility for the publication. (Art. 50(4)) | 2 August 2026 |
Risk
Breach of the transparency duties: fine up to EUR 15 million or 3 % of worldwide annual turnover, whichever is higher. For SMEs (99(6)) and small mid-caps (99(6a)) the lower amount applies. (Art. 99(4)(g))
Most common mistake
Reading Article 50 as a chatbot rule only. It also covers marking of generated content by providers and disclosure duties for deployers. (Art. 50(2)–(4))
Most common sanction
No published supervisory decision under this article yet. Fines under the AI Act can be imposed from 2 August 2025 (Art. 113).
Actions
- Does this apply to us: list every system that talks to people or generates content.
- What must we do: add a first-contact notice and machine-readable marking of output.
- What does a mistake cost: up to EUR 15 million or 3 % of turnover.
Checked against the official text: 2026-10-12
The text is quoted from the official source and is not legal advice. A national court reads the language version that binds in its jurisdiction.
Back to the act · EU acts, read down to the national law
ExploreWorldAI is operated by Valkiv Ventures AB (Reg. no. 556995-1311), Kungsgatan 8, 111 43 Stockholm, Sweden. EU-hosted, with data processing assessed against the GDPR. Contact: hello@exploreworldai.com.
Machine-readable summaries for AI agents: /llms.txt and /llms-full.txt.
