Italy becomes first EU member state to pass an AI law

An article by Rocco Panetta reports that Italy is the first EU member to pass a national law that complements the EU Artificial Intelligence Act. A Senate vote on 17 September closed a process that began on 23 April 2024 when a bill was introduced, and the bill went back and forth between both houses. The law now has 28 articles and was still awaiting formal publication in the Italian Official Gazette. The author says it is too early to judge its effects.

The law sets out general principles and then pays specific attention to work, health and justice. It also covers minors' use of AI and adds criminal sanctions. For governance, the national AI authorities are the digital agency (AgID) and the cybersecurity agency (ACN). The cybersecurity agency monitors the adequacy and security of systems and has inspection powers, while the digital agency handles notifications and promotes safe use cases. An interministerial committee is to approve a national AI strategy every two years, and an observatory at the Ministry of Labour will track AI's effect on the labour market.

Specific provisions include confirmation that AI systems may run on servers outside the EU, while public e-procurement still favours suppliers keeping strategic data in Italian data centres. Employers must use AI in ways that respect workers' integrity and avoid discrimination. In health research, simplified patient information rules allow secondary use of data without a second consent when de-identified. Children under 14 need parental consent. A new offence carries a prison term for spreading AI-generated images or videos meant to mislead about authenticity and causing unjust harm.

The author advises organisations to run AI literacy training, map their AI systems, assess compliance, set governance structures and write AI-use and procurement policies.