Italy prepares for EU AI Act transparency rules as enforcement date nears

Italian companies, public institutions and media organisations are preparing for a new layer of regulatory obligations as the European Union’s landmark AI Act enters its first binding phase on August 2, when transparency requirements for artificial intelligence systems take effect across the bloc.

Reporting on the forthcoming rules, The Local Italy explained on Monday that the legislation will require AI-generated content, including text, images, audio and video, to be labelled as synthetic, with deepfakes carrying specific disclosures. Companies deploying chatbots and other AI systems that interact with humans will also need to inform users that they are interacting with a machine.

The transparency rules are the first substantive enforcement stage of the AI Act, which was approved by the European Parliament in 2024 and is being phased in over several years. Italy, as a founding EU member, must transpose the regulation into national law through an implementing decree, which the government has not yet published but which is expected before the August deadline.

For Italy, the enforcement date presents both compliance challenges and potential competitive advantages. The Italian digital economy, though smaller than those of Germany or France, includes a rapidly expanding artificial intelligence sector, particularly in Milan and Turin. Start-ups and established firms alike will need to redesign products and services to meet the labelling and disclosure requirements, which apply to any AI system used within the EU regardless of where the developer is based.

The AI Act categorises applications by risk, from minimal to unacceptable. Transparency requirements apply primarily to systems categorised as limited risk, such as chatbots and content generators. Italian media organisations, many of which have begun using AI tools for news production and social media content, will be required to clearly mark any material that is wholly or substantially generated by AI. Failure to comply can result in fines of up to 7 percent of global annual turnover for the most serious violations, though smaller penalties apply to transparency failures.

The Local Italy’s report noted that the rules also include provisions for watermarking, a technical measure that embeds a machine-readable identifier in AI-generated content. Italian technology firms, particularly those working in generative AI, are expected to adopt watermarking systems to ensure compliance.

Italy’s data protection authority, the Garante per la Protezione dei Dati Personali, has been active in enforcing prior EU digital regulations, including the GDPR. The authority is expected to play a leading role in enforcing the AI Act’s transparency provisions, and has already signalled its readiness to coordinate with the newly created European AI Office in Brussels.

The broader context is one of regulatory divergence. The United States has taken a lighter-touch approach to AI regulation under the Trump administration, while the EU continues to assert its model of precautionary governance. Italy, which often finds itself caught between US and EU regulatory spheres, must now navigate both. Italian multinationals operating in the US market face differing requirements, complicating compliance strategies.

The AI Act is also expected to affect how Italian public institutions deploy AI, such as in administrative decision-making, healthcare triage, and police surveillance. Civil society groups in Italy have raised concerns about algorithmic bias and transparency, and the new rules are likely to spur public debate about the boundaries of automated governance.

Rome-based consumer associations and digital rights organisations have welcomed the transparency rules as a step toward accountability. However, some Italian business lobbies have argued that the regulatory burden could slow innovation and place European companies at a disadvantage compared to US and Chinese rivals. The government has not yet taken a public position on these trade-offs, but is expected to issue an implementation decree in the days ahead.