Guidelines on the implementation of the transparency obligations for certain AI systems under Article 50 of the AI Act
Commission guidelines (C(2026) 5054 final, adopted 20 July 2026) interpreting the AI Act Article 50 transparency obligations that apply from 2 August 2026, including the duty to disclose to users that they are interacting with an AI system. The 51-page document sets out when the 'obvious to a reasonable person' exception fails, what the disclosure must look like for vulnerable users, and when a one-time notice is insufficient. Conversational and companion AI receive specific treatment: sustained or emotionally sensitive interactions are named as contexts requiring periodic, context-aware re-disclosure.
Publisher
European Commission (EU AI Office)
Published
20 Jul 2026
Added
1 week ago
DOI
—
Key Findings
- Periodic reminders and context-aware disclosures are 'likely to be necessary' in riskier contexts, explicitly including interactions with children and the elderly, sustained immersive interactions involving emotional distress or addiction-like behaviour risk, and systems posing a risk of 'forming emotional attachments or dependencies (e.g. AI companions)'
- Where a system may interact with children, AI-status notifications must be child-friendly, age-appropriate, easy to understand, and accessible
- Providers must design systems to disclose their AI nature whenever asked about it by the user, and the 'obviousness' exception is read narrowly — an AI-powered robotic companion pet mimicking human-pet interaction does not qualify
- The vulnerable-user provisions apply whenever a system is reasonably likely to interact with those groups, including any system intended for the general public
Methodology Notes
Commission communication (soft-law guidance, not itself binding legislation) adopted Brussels, 20.7.2026 as C(2026) 5054 final; interprets Regulation (EU) 2024/1689 Article 50 obligations applicable from 2 August 2026. Full PDF (51 pages) downloaded from the Commission library page and provisions verified directly from the text. A separate Code of Practice on Transparency of AI-generated Content (published 10 June 2026) accompanies the Article 50(2)/(4) marking obligations.
Sources
European Commission library page (guidelines + annex PDFs) (primary)
Guidelines annex PDF (C(2026) 5054 final) (20 Jul 2026)
Archived snapshot (Wayback Machine) — preserved against link rot
Topics
Across NOPE's trackers
Tags
Cite This
APA
European Commission (EU AI Office) (2026). Guidelines on the implementation of the transparency obligations for certain AI systems under Article 50 of the AI Act. European Commission (EU AI Office). https://digital-strategy.ec.europa.eu/en/library/guidelines-transparency-obligations-providers-and-deployers-ai-systems
Related Insights
General-Purpose AI Code of Practice (EU AI Act, Articles 53 and 55)
European Commission (EU AI Office) · 10 Jul 2025
Toward Meaningful Transparency for AI Chatbots: Disclosing Persuasive Intent Reduces Persuasion
arXiv (National Taiwan University; University of Bamberg) · 12 Aug 2026