Rapporteur : Alberto CIRIO (IT/EPP), President of the Piedmont Region Reference document : Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions – AI Continent Action Plan COM(2025) 165 final

POLICY RECOMMENDATIONS

THE EUROPEAN COMMITTEE OF THE REGIONS (CoR)

General observations

1.

notes that a window of opportunity has opened for the European Union and its territories to address deep local and regional imbalances in innovation, technology, infrastructure and productivity, and to revitalise their manufacturing and industrial potential. In this context, local and regional authorities (LRAs) are best placed to drive innovation and can serve as users and facilitators of AI products and services to be the protagonists in shaping this new industrial revolution;

2.

notes that LRAs are central to the adoption of AI: as the level closest to citizens and as providers of a wide range of services, they can help to advance responsible development and the use of AI technologies across key sectors such as healthcare, transport, urban planning, energy efficiency, rural development, environment and climate, agri-food, security and defence, aerospace, entrepreneurship, education – including media, digital and AI skills – and culture, and public administration;

3.

reiterates the need to avoid the ‘AI gap’ (1) in order for it to be possible to create local, regional, cross-border, and transnational AI ecosystems in the EU, and underlines the essential role of awareness-raising actions in the design of AI solutions, and in the training and finetuning, testing and deployment of AI EU models;

AI factories and gigafactories

4.

underlines the key role of local and regional authorities in identifying the most suitable sites for the establishment of AI factories and gigafactories: Member States alone will not be able to connect AI gigafactories with local R&D centres, universities and innovation stakeholders;

5.

stresses that artificial intelligence must remain human-centred and that meaningful human oversight and final accountability for public decisions cannot be delegated to algorithms; technology must serve people, not the other way around; public value and residents’ needs must guide choices;

6.

recommends that, when deciding on the location of AI gigafactories, the following factors should be taken into account: — availability of energy sources and grids; — sustainability criteria, for instance for water and energy use; — skilled workforce; — adequate housing supply; — connection to research institutes; — the need to avoid overconcentration of critical technologies in one spot, preventing ‘overclustering’ and reducing the potential for kinetic attacks, and the threat of cyberattacks through joint nodes; — preventing a new digital divide, by ensuring high-speed and high-quality connectivity; — territorial impact assessments;

7.

Underlines that responsible AI deployment in the public sector requires an integrated approach that addresses, in coherence, ethics, privacy and data protection, information security, digital literacy and governance, rather than siloed initiatives;

8.

highlights the urgency of building new AI factories, AI gigafactories and data centres, but points to challenges in permitting procedures; underlines, in this light, the key role of subnational public authorities in permitting procedures; invites the Commission and Member States to set-up and operationalise ‘pre-permitting consultations’ to accelerate the deployment of AI infrastructure;

Data management and European AI models

9.

encourages public authorities to run an inventory of the data they collect, such as data from traffic light systems, waste collection, energy consumption, and permits and zoning for housing. LRAs could cooperate and share best practices on how to identify, store, archive and manage their data, with the goal of improving their interoperability trustworthy solutions, reuse conditions and data ownership and the cross-border interaction of public administrations at all levels; in fact, they can actively collaborate in the implementation phase of the AI Continent Action Plan and contribute to the definition of EU models and AI methodologies for regional and local development;

10.

acknowledges the need to develop ‘smaller’ algorithms at subnational level, tailored to the specific challenges of cities, SMEs or groups of citizens, but stresses that it is of the utmost importance to create EU-owned large language models, based on existing European data, aligned with EU values based on democracy and human rights; in this regard, welcomes the upcoming Data Union Strategy that will make it easier for businesses and other stakeholders to comply with data rules and ensure that European data are protected internationally; takes account of the withdrawal of the proposed AI Liability Directive, and underlines the need to find the right balance at EU level between legitimate concerns on liability risks, digital accountability, consumer protection and fostering innovation and designing rules that can be actually implemented;

11.

stresses the importance of AI predictive models tried and tested at the local and regional level which could have a strong and beneficial impact in areas such as health, urban security and mobility, energy distribution and climate adaptation and resilience, such as rainfall and hydrological risk, supporting proactive water management and flood prevention;

12.

welcomes the upgrading of European Data Spaces into Data Labs, as integral components of the AI factories, to enable the provision, pooling, and secure sharing of high-quality data across sectors; supports the large-scale implementation of the EU Data Spaces, including the Data Space for Smart and Sustainable Cities and Communities (DS4SSCC); draws attention to the tremendous amount of place-based data held by cities and regions that must be activated and mobilised to train more strongly AI models and benefit other LRAs; underlines that data privacy must be safeguarded at all levels; for this to be successful, EU investments are needed in data infrastructures and local data platforms that secure data sharing, interoperability and combined analysis between different data sources;

13.

with regard to support for creating own AI models and solutions at EU level, welcomes the recent launch of the ‘AI-on-Demand Platform’, offering a researcher-focused suite of datasets, tools, and computing resources and enabling SMEs, businesses and public sector organisations to access trusted tools, resources and ready-to-use AI modules tailored for industry needs; suggests that this platform be more closely linked to existing tools such as the Public Sector Tech Watch, online resources on the living-in.EU website and the AI Act Service Desk;

14.

to enable responsible and scalable AI adoption by regional and local authorities, recommends supporting open-source AI solutions that can be locally adapted and governed, reducing vendor lock-in and increasing transparency;

AI ecosystems and alliance building

15.

welcomes the plan’s support for AI ecosystems as these encourage mutually reinforcing cooperation between the public, private and academic sectors; notes that AI factories are intended to become larger ecosystems that will integrate AI-optimised supercomputers; stresses that local AI factories can serve as ‘beacons’ that attract innovation and anchor AI ecosystems in local economies. LRAs are vitally important in enabling AI projects, as they are the ones really engaged in trying to bring projects to their cities and regions;

16.

suggests that the Apply AI Strategy should also support smaller ecosystems tailored to local and regional needs, as the latter are truly beneficial in forging a direct link between the local private sector, social partners, smaller universities and public administration, as well as involving active citizens; welcomes, in this regard, the support for the establishment of AI antennas to provide access to supercomputers for remote and rural areas too; points out that a fast, high-quality internet connection with low latency is a prerequisite to fully benefit from AI antennas;

17.

reiterates the key role of decentralised initiatives such as EDIHs and living-in.EU, which are closest to people on the ground, in providing immediate hands-on expertise and thus contributing to building public trust in digital solutions; welcomes the second call for renewal and upgrade of the role of EDIHs, which shall increase their skills and training services to ensure the continuous learning of workers; notes that in order to sufficiently cover the increased demand for such a skilled workforce even for smaller municipalities and SMEs, the network of EDIHs will need to be substantially boosted;

18.

believes that the efficiency and quality of public services must be the prime criterion for adopting AI solutions; following the CoR’s long-standing calls for a territorial approach to EU policies, reiterates that AI adoption needs to follow the same logic of putting places and territories at the forefront;

19.

underlines that the main challenge is often to move from pilot to actual operations; urges the Commission to facilitate the adoption of ‘AI made in Europe’ by supporting pilot actions in public administrations, to ensure that local and regional pilot projects are translated into sustainable and scalable solutions; underlines that a particular focus should be given on the integration of generative AI to enhance the efficiency, effectiveness, and accessibility of public services for citizens at all levels;

20.

stresses the essential role of procurement and supply management in ensuring support for AI solutions developed in the EU by European experts and in enabling LRAs to adopt safe AI solutions; calls for a minimum set of contractual safeguards in public procurement, such as transparency on model training, enforceable data-ownership, open standards, and independent audit/inspection rights; notes that exchanges of practice between public authorities have already begun – a set of guidelines and templates for contractual clauses for public procurement of different digital solutions, developed under the living-in.EU movement, is now publicly available;

21.

welcomes the revival of the AI Alliance and expresses the CoR’s readiness to relay public sector experience to its discussions as LRA representatives.

AI as a strategic technology

22.

underlines the strategic nature of AI as a new technology: the EU needs to ensure that AI talent is developed in the EU and that industrial property is not transferred to third countries; invites the Commission to come up with innovative solutions for retaining AI knowledge in the EU, such as prohibiting participation in public tenders and EU programmes by those companies that violate the non-transferability of AI solutions to third countries or third parties, and highlights, more broadly, the EU’s efforts on the screening of foreign investment and the CoR’s work in this regard (2);

23.

points out that account must also be taken of the positive response from, and efforts by, third country providers to meet Europe’s requirements and comply with EU rules. The local level should also be able to assess actual compliance with commitments and potential loopholes. Planning and support are therefore required not only in the event of intentional security risks and worst-case scenarios, but also in situations where a third-country provider tries to comply with the rules and act as a trusted partner. The EU must also clearly state its position in relation to such cases;

24.

stresses the need to ensure that EU-based AI models, solutions, start-ups and scale-ups can grow and be widely used in the EU and that they are not subsequently taken over or bought by third countries’ entities, in a manner that may pose significant security and digital sovereignty risks;

25.

calls for a more joined-up approach across all government levels in the EU, to the deployment of 5G infrastructure and networks in the EU; this is a key enabler of artificial intelligence systems, as it can provide real-time data collection and analysis and prepare the actions and investment for 6G. The Committee also calls for clear incentives to truly mobilise private capital;

26.

warns against growing dependence and reliance on third countries in relation to advanced AI chips, where the EU may create new dependencies; highlights the importance of the European Semiconductor Regions Alliance, which brings together front-running regions to support the strategic planning of investments to produce new generations of semiconductors, by mobilising and enhancing regional capacities and innovating ecosystems; in this connection, welcomes the expected review of the EU Chips Act and invites the Commission to factor in territorial specificities and access to critical raw materials, a skilled workforce and energy supply; invites the Commission to conduct a Territorial Impact Assessment ahead of the EU Chips Act review and offers the CoR’s experience in this regard;

27.

calls on the European Commission to involve the CoR in in its self-described ‘Team Europe’ approach to harnessing potential through the EU’s Global Gateway (3), and emphasises the CoR’s capacity to raise greater awareness of the gateway in the Member States;

28.

welcomes the recent publication of the Quantum Europe Strategy, which sets out actions to ensure that the EU can become a quantum powerhouse and remains competitive on a global scale; reiterates the need for alignment between the Quantum Europe Strategy and the AI Continent Action Plan in order to develop combined projects in AI and quantum technologies; insists that AI factories, AI gigafactories and data labs be designed to accommodate quantum computers as soon as possible after their commercialisation; this will have an impact not only on AI solutions as such, but on cryptography and data security in general;

29.

notes that cybersecurity could have been covered more extensively by the AI Continent Action Plan, especially in view of the huge impact AI – potentially in combination with the quantum breakthrough – can have in this field, in order to identify and predict threats and vulnerabilities; adds that cybersecurity of and using AI should also include ensuring its trustworthiness, for example by making sure that it has human oversight and is robust enough to counter cyberattacks; against this backdrop, calls for the exception provided for in the AI Act for the area of security to be promptly evaluated in view of the goal of a digitally sovereign EU that can counter digital threats;

30.

reiterates its recent call for an increased presence of IT control systems and proper training of operators to protect against cyberattacks, manipulation and interference by state and non-state actors (4); emphasises that AI applications can be very useful in predicting, detecting, responding to and combating misinformation and disinformation campaigns, in particular following major unexpected events with the potential to disrupt critical infrastructure, for example after extreme weather events and natural disasters where local and regional authorities are often frontline actors;

Support and guidance for public administration

31.

underscores that LRAs are pioneers in cross-border and transnational collaboration with each other in the development of public sector AI solutions, and are actively seeking interaction with other levels of government;

32.

stresses that more support is required from the national and EU levels to provide guidance and assistance for local and regional administrations on AI compliance, in particular regarding liability, privacy and data protection; urges EU and national authorities to provide LRAs with sufficient resources and concrete support for navigating and complying with EU legislation, making it more accessible;

33.

underlines the need to roll out guidelines for cities and regions on AI Act implementation; and related regulatory frameworks, such as the European Health Data Space (EHDS); welcomes, in this light, the establishment of AI Act Service Desks and calls for local and regional administrations to be treated as a priority group, providing them with tailored information, also with practical toolkits and templates on modalities of their functioning and stresses that it is essential for these desks to operate in all EU languages so that they are accessible to all; reiterates that an AI chatbot can complement but not replace the role of human experts;

34.

welcomes the forthcoming digital Simplification Omnibus as an opportunity to strengthen connections with smaller and scaling companies notes that the regulatory burden continues to weigh heavily on EU companies; and stresses the need for truly innovation-friendly regulations that encourage investment and risk-taking, translating research and development into marketable technologies;

35.

draws attention to the CoR’s proposal (5) to set up a mechanism for the collection and exchange of strategies and guidelines on the use of AI at local and regional level, and supports the idea of EDIHs being upgraded into AI Experience Centres to provide enlarged support; notes the existence of the Public Tech Watch, which collects user cases and solutions from across the EU, and suggests that it be upgraded into a fully-fledged best practices repository, with a user-friendly interface, easy filtering and easy-to-understand explanations, translated into all EU languages; welcomes in this regard the work done by the Joint Research Centre to classify the cases collected and create a typology;

36.

acknowledges that AI in the public sector will be a priority for the AI Office and calls for greater and more systematic cooperation between the Commission and the CoR;

Regulatory sandboxes and monitoring

37.

welcomes the support for the creation of regulatory sandboxes where AI solutions can be safely tested before being widely deployed, in particular in connection with data use and privacy requirements, which are a cause for concern for many frontrunning LRAs; reiterates the recent CoR call to support interregional networking of sandboxes (6) that can share experiences and best practices, ensuring gradual convergence and innovation compatibility across regions and cities, to ensure easy access for SMEs and startups in particular, and to support the development of dedicated EU urban sandboxes so that cities can test AI and GenAI solutions safely;

38.

welcomes the AI Office’s proposal to set up an observatory to monitor developments in AI and its implementation, and calls for the CoR to be more involved in monitoring; in this context, reiterates the CoR’s readiness to contribute to the work of the EDIHs through regular stakeholder consultations via its RegHub network; recommends establishing relevant indicators to monitor AI adoption and suggests that ‘the number of AI solutions in public administration’ be included as a new indicator in the Digital Decade reporting;

39.

in line with findings from the CoR’s recent study on AI and GenAI adoption by local and regional administrations, requests that the Commission outline standardised monitoring and Key Performance Indicators to track AI’s impact, adjust strategies and align with technological advances;

40.

recommends again the use of proper adjusted indicators to monitor different digital gaps at EU level; notes in this light the following ongoing initiatives: 1) LORDIMAS, an application that features close to 400 local and regional entries and has been created within the framework of the living-in.EU movement; 2) ESPON DIGIREG, with the objective of mapping existing digital divides within and across European regions (to NUTS3 level). The Committee calls for outcome-based metrics such as investment levels, adoption rates, skills development and data-sharing progress to be applied;

41.

underlines its call (7) for the creation of a European conformity certification system for AI tools used in EU public administrations to ensure their compliance with EU legislation; takes note of the new tasks that the AI Office is intended to assume and proposes that existing networks and alliances be used to implement it: decentralised management could prove to be a cheaper, more efficient solution than centralised and top-down approaches;

Training and skills development

42.

welcomes the Commission’s actions on training and skills development to keep up with the increased need for an AI-skilled workforce, and fully supports the AI Skills Academy initiative; in light of the continuing importance of teaching AI skills locally; reiterates the importance of avoiding algorithmic bias in AI-related technologies and calls for increasing girls’ and women’s participation in the field of science, technology, engineering and mathematics (STEM); reiterates furthermore the CoR’s recent call to explore innovative solutions to attract and retain talent in the public sector, underlining the importance of creating high-quality jobs;

43.

welcomes the Action Plan’s recognition of social dialogue as key to anticipating and addressing skills needs, as well as to facilitating the adoption of digital and AI technologies in the workplace in a fair and inclusive way;

44.

points out that, according to the aforementioned CoR study (8), the vast majority (86 %) of local and regional representatives responded that AI is either only used by individual employees on a non-systematic basis, only partially used in supporting specific functions or departments, or finding limited use in pilots; At the same time, AI must be introduced in a way that is geared to needs; therefore hopes for tailored and concrete support for the local and regional level and for small municipalities and rural and remote areas that may lack in-house expertise in AI governance, procurement and ethics, ensuring that no one is left behind in the AI transition;

45.

stresses that public-sector AI requires iterative cycles of design, deployment, evaluation and adaptation with feedback loops from users and residents; monitoring should focus on outcomes and public value rather than inputs alone;

46.

points out that LRAs have an important role to play in increasing digital and AI literacy both in the administration, among politicians and decision makers, and among the general public, for example by organising citizen engagement hubs and panels, social AI labs, and other solutions;

47.

also promotes the creation of public-private partnerships that bring together universities, research centres, and technology companies which could provide for shared learning programmes, on-the-job experience, and practical projects, giving staff real experience with AI in public services;

48.

also highlights that strengthening skills and enhancing AI literacy, starting in primary education and continuing through higher education with the close cooperation of the private sector, could help to reduce dependence on external providers. In fact, while Europe should strive to match the US in innovative potential, it should aim to exceed it in providing opportunities for education and lifelong learning – ensuring that the benefits of AI are widely shared and any negative impacts on social inclusion are minimised;

Funding and financial support

49.

takes note of the budget available in the EU Competitiveness Fund and the number of current and future support actions provided by the Commission, especially for regional public authorities and small companies; however, suggests that more light should be shed on different financing opportunities – under shared management and direct and indirect management – in terms of how they operate, their potential combinations, and consortia building;

50.

calls on the Commission to establish a single ‘AI Fund for regional and local authorities’, with the aim of accelerating the full implementation of all the actions envisaged in the AI Continent Action Plan;

51.

welcomes the TSI projects, which help to raise awareness among cities and municipalities about the digital transformation; calls on the Commission to replicate such projects, taking into account the regional specificities of their beneficiaries;

52.

asks Member States and the Commission to increase the EU and national funding resources available to finance AI projects; seeking to ensure that they reach all authorities, including the smallest ones; welcomes the recent GenAI4PA initiative to create consortia to implement pilot AI projects in public administrations, as this will set an example of successful AI solutions that can be replicated shared and scaled across all EU Member States and regions;

53.

calls upon the Commission to keep the Horizon funds and the Competitiveness Fund accessible for LRAs, under the new MFF 2028-2034;

54.

highlights the crucial role of private capital: the EU cannot become an AI leader with public funding only; there needs to be a greater move to embrace equity finance and stronger synergies between financing at all levels, and the Commission and Member States should strive to find a suitable financing mix to support AI projects – once again, EDIHs can also enable matchmaking between public and private capital and should be supported by the EU in the long term; moreover, supports the expansion of public-private partnerships, such as the Innovative Health Initiative, and models such as outcome-based investment, co-investment agreements, regional AI investment platforms and innovation pacts.

---

(1)  Opinion of the European Committee of the Regions – Challenges and opportunities of artificial intelligence in the public sector: defining the role of regional and local authorities (OJ C, C/2025/288, 24.1.2025, ELI: http://data.europa.eu/eli/C/2025/288/oj).

(2)  Opinion of the European Committee of the Regions – Proposal for a Regulation on the screening of foreign investments in the EU (OJ C, C/2025/290, 24.1.2025, ELI: http://data.europa.eu/eli/C/2025/290/oj).

(3)   https://international-partnerships.ec.europa.eu/policies/global-gateway_en.

(4)   OJ C, C/2025/288, 24.1.2025, ELI: http://data.europa.eu/eli/C/2025/288/oj.

(5)   OJ C, C/2025/288, 24.1.2025, ELI: http://data.europa.eu/eli/C/2025/288/oj.

(6)   OJ C, C/2025/288, 24.1.2025, ELI: http://data.europa.eu/eli/C/2025/288/oj.

(7)   OJ C, C/2025/288, 24.1.2025, ELI: http://data.europa.eu/eli/C/2025/288/oj.

(8)  CoR study AI and GenAI adoption by local and regional administrations, https://op.europa.eu/en/publication-detail/-/publication/40363d58-bdc8-11ef-91ed-01aa75ed71a1/language-en?WT.mc_id=Searchresult&WT.ria_c=125603&WT.ria_f=8099&WT.ria_ev=search&WT.URL=https%3A%2F%2Fcor.europa.eu%2F.