The European Commission adopted Commission Guidelines C(2026) 5054 final on 20 July 2026, implementing the transparency obligations of Article 50 of Regulation (EU) 2024/1689, the AI Act. The obligations entered into application on 2 August 2026. The guidelines are final interpretive guidance. They bind national market surveillance authorities and the AI Office without requiring separate national transposition.
Article 50(1) of the AI Act requires providers of AI systems that interact directly with natural persons to inform users they are communicating with AI, unless that is obvious from context. Article 50(2) requires providers of AI systems that generate or manipulate images, audio, video, or text to ensure outputs carry machine-readable marks enabling automated detection of AI origin. Article 50(3) requires deployers of deepfake systems to disclose the artificial nature of depicted content to affected persons. National market surveillance authorities hold primary enforcement jurisdiction in their territories. The AI Office enforces against general-purpose AI model providers. The European Data Protection Supervisor enforces when EU institutions act as providers or deployers under Article 74(6) of the AI Act.
AI system providers releasing chatbots and virtual assistants to EU users must update product interfaces to carry real-time disclosure before any user engagement begins. Providers of generative video, audio, and image tools must embed C2PA-compliant or equivalent machine-readable content credentials in all AI-generated outputs distributed in the EU. Deployers using AI to generate deepfakes in commercial contexts must attach visible disclosure overlays to each output. Non-compliance with Article 50 obligations carries penalties under Article 99 of the AI Act of up to EUR 15,000,000 or 3 percent of worldwide annual turnover, whichever is higher.
Article 50(3) provides a carve-out for deepfake disclosure where the depicted person has explicitly authorised the content or where the system is used solely for personal, non-commercial purposes. The guidelines identify open technical questions on machine-readable content credential standards and point to the C2PA specification as an interim reference pending formal harmonised standards under the AI Act. Several EU member states had not yet designated national market surveillance authorities for AI Act enforcement as of 2 August 2026, creating temporary gaps in national-level enforcement coverage.
Licentium advises AI system providers and deployers on EU AI Act compliance obligations across the product development and release lifecycle. Work we undertake includes Article 50 transparency obligation gap analysis, content credential implementation review, AI governance program design, cross-border AI product regulatory mapping, and engagement with the AI Office.