EU begins enforcing Article 50 transparency rules for generative AI

The European Commission has started enforcing the EU AI Act's transparency obligations under Article 50, activating new labelling and disclosure requirements for chatbots, synthetic media and other AI-generated content, the Commission said.

On 2 August the Commission said enforcement will be carried out through the European AI Office together with national market surveillance authorities and that public guidance and a Code of Practice on marking AI-generated content have been issued to support compliance. The Commission also said the European AI Office has expanded its enforcement staffing and evaluation capacity to carry out checks and coordinate oversight across member states.

Article 50 requires that certain AI systems make clear to users when they are interacting with or consuming content produced by AI. The Commission's move makes those transparency obligations operational for a range of consumer- and business-facing generative AI tools, including conversational agents and synthetic audiovisual material, according to the Commission's announcement and explanatory guidance highlighted by AI-Jarvis.

For employers and HR teams the change is immediate: chatbots used in candidate communication, AI-generated internal content and marketing materials distributed in the EU now fall within a legal framework that obliges affirmative disclosure. That means HR and communications managers who deploy third-party or in‑house generative models will need to check whether their vendor contracts, user interfaces and published materials meet the new marking expectations set out by the Commission and the Code of Practice.

The European AI Office will lead coordination with national market surveillance authorities, the Commission said, signalling a two‑tier approach in which EU-level supervision is paired with on-the-ground checks by member-state agencies. The Code of Practice and accompanying guidance are positioned as practical tools to help developers, platforms and downstream users interpret how to label AI outputs in different contexts.

The enforcement launch reflects a broader regulatory push across jurisdictions to make AI usage more auditable and transparent. Regulators in the EU have been explicit that transparency obligations are central to trust in generative systems, and the operationalisation of Article 50 places the bloc among the first to turn a comprehensive AI law into day‑to‑day obligations for businesses that use or publish AI content.

What the Commission did not disclose is how enforcement priorities will be set or how sanctioning will play out in workplace contexts. The Commission and the European AI Office have not published detailed examples showing how disclosures should look inside candidate‑facing chat flows or internal HR communications, nor have they outlined a timetable for inspections or the thresholds that would trigger penalties. That leaves employers and suppliers to interpret the Code of Practice without a catalogue of enforcement precedents.

The move marks a new phase in which generative AI is not only a technology decision for HR and communications teams but a compliance issue that will be scrutinised by regulators. As the European AI Office and national authorities begin checks, HR leaders in the EU will need to reconcile their use of conversational agents and synthetic content with a legal requirement to label AI outputs — a shift that could change vendor evaluations, candidate‑communication workflows and the way organisations document their use of generative systems.