The European Union’s AI Act will make it mandatory from 2 August to label all content generated by artificial intelligence, marking the first major enforcement step since the legislation was formally adopted two years ago. Article 50 of the Act, which becomes applicable on that date, establishes the right of citizens to know whether they are interacting with a chatbot, a virtual assistant or a human, and whether a text, video or image has been artificially created. The provision ends what Italian journalist Francesco Giorgino, writing in La Gazzetta del Mezzogiorno, called an era in which “ambiguity is no longer tolerable.”

Google has publicly declared its support for the accompanying Code of Practice on Transparency of AI-Generated Content, a voluntary framework designed to help companies meet the Act’s obligations. The company said it is deploying SynthID, a digital watermarking system that embeds imperceptible identifiers into AI-produced images, video, audio and text, and is also backing the Coalition for Content Provenance and Authenticity (C2PA) standard, which allows platforms to display the origin of a piece of content. The European Commission has described the code as a practical guide for firms, while stressing that participation remains voluntary.

Viewed from Rome, the transparency drive has already produced a political ripple effect. Prime Minister Giorgia Meloni and Elly Schlein, leader of the opposition Democratic Party, have struck what Italian media termed a “digital non-belligerence pact,” pledging that their parties will not use AI to attribute invented facts or opinions to opponents during the upcoming election campaign. Meloni had earlier warned that deepfakes are dangerous tools capable of manipulating reality and interfering in democratic processes, citing a fake image of herself that circulated on social media. Giorgino called on all political forces to join the agreement, arguing that in an era of polarisation it is essential to show citizens who creates, approves and controls political content.

The AI Act’s transparency requirements are the first binding rules of their kind in a major jurisdiction. While the code of practice is not mandatory, companies that follow it gain a clearer path to compliance. The European Commission expects the labelling system to improve digital literacy and reduce the spread of misinformation. The next concrete step is the 2 August deadline, after which AI-generated content circulating in the EU must be marked as such.