Deepfake and politics, how the rules are in Italia and Europe. Alarm after the latest episodes
From the AI-modified photo of the agent in Turin to the Lithuanian case of the video dubbed with synthetic voices, manipulated content enters the political confrontation
by Lorenzo Pace (Il Sole 24 Ore) and Ieva Kniukstiené (Elta, Lithuania)
We could say that the debate on the use of artificial intelligence in Italia has entered a second phase. No longer theoretical, thus limited to hypothetical cases, but to concrete episodes. The latest is recent, from the end of January, and concerns a photograph. Namely that of the officer injured in the clashes during the pro-Askatasuna demonstration in Turin, published directly by the police. The problem? It was modified with Ai, even if not blatantly.
To recognise the artificial 'hand', in fact, the video of the attack was decisive, from which fact-checking sites noted certain details and made it clear that the image was not authentic. The issue had already become political, like the whole affair, in the days when, among other things, the final draft of the government's security decree, approved a few days later in the Council of Ministers, with measures also in favour of the agents, was being processed.
The Ai touch involves photos, videos and even audio. As happened with the use of a synthetic voice, cloned with artificial intelligence techniques, attributed to Defence Minister Guido Crosetto.
On the regulatory side, the European Union has initiated a structured response with the Ai Act, which introduces transparency obligations for artificially generated content and stricter rules for high-risk uses. However, the entry into force of the regulation will be gradual and not immediate. In the meantime, the Digital Services Act mainly focuses on the responsibilities of online platforms in the moderation of content, leaving the issue of intentional political use of synthetic media by parties, candidates or organised groups partly uncovered.
This is why there have been other initiatives in Italia. Pagella Politica and Facta, for instance, have proposed a public commitment to parties not to use artificial intelligence to create or disseminate misleading content during election campaigns, inviting them to explicitly declare any use of Ai tools in political communication. Although the proposal has no binding effect, it is intended to temporarily fill the regulatory gap and strengthen voter confidence, relying more on reputational pressure than on formal sanctions.
