Source: Assostampa Sicilia

The application of Article 50 of the European AI Act began on 2 August 2026, establishing transparency requirements for chatbot interactions, virtual assistants, and synthetic content. While enforcement of other provisions regarding high-risk AI systems has been postponed to 2027 and 2028, the current rules require that the public be informed when engaging with machine-generated content. These obligations directly affect public administrations, press offices, communication departments, URP, and web editorial teams, which must organize their use of AI-based tools accordingly.

For content of general interest generated or supported by artificial intelligence, the AI Act clarifies that it does not require automatic labeling if subject to effective human editorial control. This control involves rigorous fact-checking, data verification, editorial oversight, assuming full responsibility for published materials, and ethical evaluation of language and social context. The FNSI highlights the need to regulate AI use within employment contracts to ensure that algorithms remain tools supporting journalist work without replacing their professional responsibility. Additionally, public entities must maintain inventories of AI tools, specify data not processable by AI, and provide continuous role-specific training to staff.