The road to the AI Act: the Italian approach - Part 1: the Italian Strategy on AI 2024-2026

April 23, 2024

On 13 March 2024, the European Parliament endorsed the AI Act with an overwhelming majority: 523 votes in favor, 46 against and 49 abstentions. The adopted text implements the provisional political agreement reached on 8 December 2023 by European institutions (see our summary here), and it is the result of almost three years of discussions (the first version of the AI Act was indeed published by the European Commission on 21 April 2021).

The AI Act is still subject to a final lawyer-linguist check and also needs to be formally endorsed by the Council. It will apply in all EU member states 24 months after its entry into force, except for (i) bans on prohibited practices, which will apply six months from the law's entry into force, (ii) rules on general-purpose AI models, governance, and penalties, which will apply 12 months from its entry into force, and (iii) obligations for high-risk systems, which will apply 36 months from the entry into force. Specific rules will describe cases where the AI Act will apply or not to those AI systems already placed on the market or put into service before the date of application of the AI Act.

As there is still quite a long way to go, many initiatives are flourishing in anticipation of the AI Act. For example, even before the AI Act was final the European Commission established an AI office and published its governance strategy with the explicit aim of setting a positive example by working on implementing the AI Act's requirements way ahead of the legal deadline.

In addition to European Institutions, member states are also taking action.

As far as Italy is concerned, two initiatives in particular are worth mentioning: the launch of the Italian Strategy on AI for 2024-2026, and the draft law on artificial intelligence.

The Italian Strategy for AI for 2024-2026 has been prepared by the Agency for Digital Italy (AgID) and the Department for Digital Transformation of the Presidency of the Council of Ministers. At the date of this piece, the contents of the Italian Strategy have been anticipated by the (leaked!) publication of an executive summary.

According to the Executive Summary, the Italian Strategy on AI for 2024-2026 will develop along four main sectors: Scientific Research, Public Administration, Enterprises and Education.

Each sector has specific objectives and is focused on specific actions, for example:

- Scientific Research:

• Consolidate the Italian research ecosystem, including by promoting the collaboration and exchange of competencies and knowledge among universities, research centers and ICT companies that develop or use AI systems.
• Retain and attract talent, implementing an extraordinary plan of hiring that allows to absorb the excellence trained through the PNRR initiatives and that enables specific actions of brain gain in AI.
• Develop national large multimodal models (LMM) that abide by European values and regulations and that possibly focus on specific application domains where Italy has a strong international recognition and a clear competitive advantage in the definition of reference datasets.
• Promote interdisciplinary projects for social well-being, launching competitive initiatives aimed at centering on people and society the digital transformation enabled by AI.
• Finance research on next-gen AI via a call for ideas.
• Enhance international cooperation, also aiming at fostering the implementation of common standards.

- Public Administration:

• Define guidelines for promoting the adoption of AI in the public administration, providing a methodology to address technical, ethical and regulatory issues.
• Define guidelines for AI procurement in the public administration, integrating the existing provisions for the ICT sector and supporting the public administrations in the tenders and in specific framework agreements.
• Define guidelines for the development of AI applications in the public administration, ensuring compliance with national and European regulations, consistency with national strategies and development of training initiatives for staff.
• Simplify the public administration for citizens and businesses, identifying available or tailor-made solutions suitable to meet the functional and operational requirements of the public administration.
• Improve processes in the public administration, developing at least three national projects on areas to be identified also in light of the impacts and risks of the AI systems concerned.
• Promote AI literacy in the public administration, pairing innovation projects with upskilling programs.

- Enterprises:

• Create an ecosystem of facilitators for AI in SMEs that intercept the innovation needs of enterprises, providing innovation services based on AI and enabling interoperable solutions also in the context of supply chains.
• Support the development and adoption of AI solutions, providing a dedicated financial fund for specific projects promoted by networks of enterprises or by single enterprises, accompanied by a plurality of other systems such as corporate venture, private equity and vouchers for innovation.
• Establish a network of laboratories for the development of AI applications in industrial contexts, involving enterprises in collaboration with research entities, to conduct applied research and with a market focus, enabling new solutions and experimenting with verticalizations.
• Support the development of startups in AI, defining a specific fund that targets new technological enterprises while fostering the interconnection with enterprises that use AI and with entities that already operate in the acceleration and incubation paths, both at the academic level and in the private sector.
• Support ICT companies that develop AI technologies, defining measures to support the management of compliance and certification practices, and to encourage access to the sandboxes provided by the AI Act for the experimentation of innovative solutions.

- Education:

• Structure pathways for introducing AI in schools, to implement rules on education to digital citizenship, involving teachers and students across the entire national territory.
• Promote "mobility" as a training tool for AI, incentivizing internship programs in companies, internships for the exchange of skills between universities and businesses, visiting periods in research centers, and high-level apprenticeship programs.
• Spread AI literacy in universities, strengthening AI degree programs in the current landscape through specific orientation programs, and integrating courses on the basics of AI
• Support the National PhD in AI, by co-financing scholarships for at least three additional cycles and financially supporting educational initiatives.
• Activate upskilling and reskilling programs businesses and Public Administration, structured on different action plans in relation to the various roles and functions of the work environment, from technical profiles up to managerial staff.
• Educate on the use of AI tools, to encourage a conscious use of new technologies, through initiatives in the territories to be mediated by communication channels.
• Promote technical high school courses on AI, aiming at educating individuals with top-level technological competencies, involving universities and ICT companies specialized in the development of AI-based solutions.

The strategy also identifies two types of enabling actions:

• Infrastructural actions aim at creating and enhancing the resources and tools that support the development and adoption of AI solutions. They include creating a national repository of datasets and models and enhancing the network infrastructure for AI.
• Operational actions are focused on the implementation, coordination and monitoring of the national strategy. These aim at establishing a governance structure and a dedicated fund for the management and evaluation of AI initiatives also by establishing a Foundation for AI under the Presidency of the Council of Ministers.

Each business should closely follow the developments of the Italian Strategy, to avoid missing any opportunities - not only with regards to investments, but also as to initiatives fostering innovation.

According to the draft law on artificial intelligence (version dated 8 April 2024, 2:30 pm - leaked!), the Italian Strategy is to be drafted and updated by the Department of the Presidency of the Council of Ministers in charge of technological innovation and digital transition and has to be approved by the Interministerial Committee for Digital Transition; the monitoring of the implementation of the Italian Strategy will be ensured by the above Department, through AgID and the National Cybersecurity Agency (ACN).

The draft law on artificial intelligence has already raised many concerns: One revolves around the appointment of the authority in charge of supervising the implementation of the AI Act, once it is effective. As of today, both the mentioned Agencies-AgID and ACN-have been proposed as national authorities for AI, mainly due to their technical competences.

The (not yet final) choice to appoint these governmental agencies has led the Italian Data Protection Authority, which has consistently emphasized the need to have a single supervisory authority in charge of both AI and data protection matters, to issue a letter to the Italian Parliament reiterating the importance of having only one AI authority, fully independent.

To find out more, have a look at our piece on the draft law on AI and on the Italian Data Protection Authority's letter on the authority to be put in charge of the application of the provisions contained in the AI Act.

This article is updated as of 15 April 2024.