What does AI adoption mean for the workforce in social security organisations? This chapter discusses the impact of AI adoption on jobs and skills within the sector. It also explores good approaches to prepare the workforce and improve in-house AI capability through hiring and training. Drawing on examples from social security institutions and public administrations in the OECD, the chapter highlights workforce strategies that support an effective and trustworthy use of AI in line with organisational goals.
Harnessing Artificial Intelligence in Social Security
Use Cases, Governance and Workforce Readiness
Abstract
The adoption of AI in public sector organisations is not only a question of technology, but importantly, of a workforce that develops, implements, and uses AI tools. Successful AI adoption requires people who have the right skills, are aware of benefits and risks of AI, and work collaboratively on improving the way in which their institution provides value to the public.
The performance of AI tools on cognitive tasks such as information processing is constantly improving. This raises the question what changes AI adoption brings for the public sector workforce. AI will have an impact on workforce and skills needs, as it may augment human work, automate specific administrative tasks, and transform roles and organisational processes. To harness the benefits of AI effectively, responsibly, and in line with institutional goals, public institutions need to align AI implementation with strategic workforce development.
The OECD AI Principles point out the critical role of governments to “empower people to effectively use and interact with AI principles […], including by equipping them with the necessary skills”. Workforce development strategies – spanning recruitment, outsourcing, and training, should be an integral part of the institutional approach to build AI capability. Social security institutions have three main levers to do so: recruitment, outsourcing, and training. Recruitment of scarce tech talent is challenging, and while outsourcing is a common practice in government institutions, it carries risks that need to be managed.
Ultimately therefore, building in-house capability through targeted skills development and an organisational culture in favour of innovation and continuous learning may come with the benefit of greater effectiveness of AI solutions, higher accountability, and better alignment with public values. Training is an important lever across government institutions to build an AI-ready workforce capable of harnessing the benefits and managing the risks of AI. The understanding of skills in this context, following Mergel et al. (2023[1]), goes beyond technical and transversal skills, and encompasses awareness and knowledge of technological trends, policy, legal and ethical concerns, as well as management practices.
This chapter starts with a literature review of the skills and workforce implications of AI in the public sector. Second, it discusses international good practices in recruitment and outsourcing that support public administrations in incorporating the skills and building the capacity they need to adopt AI. Third, it discusses examples of training initiatives within public institutions to build AI capability, based on stakeholder interviews. Lastly, the chapter draws wider conclusions for a strategic approach to workforce development that supports AI adoption to improve public services.
As social security institutions and public administrations across the OECD adopt AI tools, there are implications for their workforce and the skills they require. There is indeed growing evidence that AI can support individuals at work to complete certain tasks faster and produce better results (Czarnitzki, Fernández and Rammer, 2023[2]; Brynjolfsson, Li and Raymond, 2023[3]). For now, AI seems to be more likely to automate tedious, repetitive tasks. Employee surveys from the private sector indicate that, as a result, both the pace of work as well as employees’ performance and enjoyment have increased (Lane, Williams and Broecke, 2023[4]). At the same time, the capabilities of AI tools continue to improve, including on performing non-routine cognitive tasks such as information-processing (OECD, 2023[5]).
Existing research on the potential impact of AI on the public sector workforce is limited (Mergel et al., 2023[1]), but the literature on AI and labour markets in general provides important insights. In the overall economy, while many jobs have a high exposure to AI, evidence of widespread job loss due to AI is so far limited (Acemoglu et al., 2022[6]; OECD, 2023[5]; Gmyrek et al., 2025[7]). Importantly, the technical feasibility of automating certain tasks may not immediately translate into real-world application, as organisations assess the costs, benefits, risks, and regulatory barriers of introducing AI. Social security institutions have high scrutiny for new technology, and they are also less flexible in re-allocating human resources compared to private companies. Nevertheless, three main conclusions can be drawn from the existing literature about the workforce impact of AI for social security organisations.
There is potential to improve public-sector productivity through AI adoption.
Social security institutions are major employers in all OECD countries, and many of them face staff shortages, high workloads, and fiscal pressures. Among their core functions are administration and support tasks, for example, to process benefit claims or collect standardised contributions. The potential impact of AI adoption to reduce the administrative burden and free up time and capacity for more interesting and higher-value tasks is therefore substantive (Mergel et al., 2023[1]; Atkinson and Browne, 2024[8]). Reallocating job roles and human resources can be challenging within the public sector and may take place over a longer period of time.
AI is indeed already applied to automate specific tasks and processes within public sector organisations. The examples in Chapter 1 of this report are insightful. A recent study commissioned by the German Federal Press (Bundesdruckerei) suggests that even quite different administrative procedures across government can be broken down into similar process components which may be enhanced through technology. Particularly, rule-based administrative processes have potential to be sped up applying data and AI solutions (Bundesdruckerei and Possible, 2025[9]).
An interesting example is the use of AI by Kela, Finland’s national social security institution, to automate the intake and handling of attachments linked to benefit applications. Their AI platform standardises a range of different types of documents through text recognition, image correction, format conversion etc. in a single workflow and saves time for administrative case workers (see also Chapter 1).
Most stakeholders noted that the full automation of certain tasks or processes is not yet possible, however, as any result or decision requires human approval. In most institutions, this is critical to clearly attribute accountability for decisions and avoid the risks posed by technology malfunction, misuse, bias, or a lack of transparency in AI use.
The challenge faced by institutions, as a result, is not only to determine where and how the application of AI makes sense, but also understanding the way in which work processes may need to change to harness the benefits of AI technology more fully. In other words, applying AI narrowly to existing tasks will not fully exploit the potential of the technology for more effective and better services.
Most institutions interviewed for this report seem to be aware of the potentially large productivity boost that AI could bring to public service and benefits provision. This potential may grow further as the capabilities of AI systems keep improving and governments build the necessary AI and data governance and infrastructure and prepare their workforce to adopt AI.
AI changes skill needs in public administration.
Technology adoption, including AI, increases the demand for specialised digital skills. As in the private sector, public institutions are facing a growing demand for specialised digital and data professionals who have advanced technical skills in areas such as data science and machine learning. To respond to these needs, workforce training and reskilling matters, and also recruitment is an important part of the response (Chinn et al., 2020[10]). This is a challenge, as many government institutions in the OECD compete against the private sector to attract and retain qualified digital and data professionals (Gouvernement Français, DINUM and INSEE, 2021[11]; Medaglia, Mikalef and Tangi, 2024[12]). At the same time, OECD research finds that while the demand for specialised AI skills is growing, it still only affects a tiny fraction of jobs (Green, 2024[13]; Borgonovi et al., 2023[14]).
Besides a small group of specialised digital and data professionals, most employees in social security institutions will not require advanced technical skills related to AI. Nevertheless, AI adoption will change the tasks these employees do, the way their work is organised and the skills required to perform their jobs (Green, 2024[13]; OECD, 2023[5]). Most employees will have to be capable to effectively and responsibly use and interact with AI tools at work, which is referred to as AI literacy in the EU’s AI Act. A key foundation for AI skills is a good baseline of basic digital skills, which in some public sector institutions is a challenge. In addition, continuous learning for all groups of staff will be necessary to gain elementary knowledge of how AI works and what its limitations are.
Staff at all levels will increasingly need a broad range of skills that are complementary to AI systems, in particular higher-level cognitive skills such as problem-solving or critical thinking, and transversal skills such as social, management, or teamwork skills (Theben, Plamenova and Freire, 2023[15]; OECD, 2023[5]). These complementary competences will not only permit AI adoption and effective use, but also enable employees to engage in continuous learning, adjust to and shape changes in the organisation of work, and find innovative ways of using AI tools to support them in achieving the objectives of their institution.
The example of an AI solution developed by ITSV, the IT service provider for social security institutions in Austria, highlights how task content and skills demand may be changing. ITSV developed an AI tool to support process claims management, which automated the administration of simple routine claims. The introduction of the AI tool had a secondary effect of leaving mostly complex cases to administrative case workers, and staff perceived that the difficulty of their work increased as a result. In this case, the value-added of the AI solution may have been partially overshadowed by changing skills requirements and increasing task difficulty for staff.
Finally, leaders and decision-makers in public institutions have a critical role to drive strategic AI adoption while maintaining accountability, managing risks, ensuring compliance, and protecting data. Good leaders are essential to ensuring the effective and responsible uptake of AI technology (Valle-Cruz, García-Contreras and Patricia Muñoz-Chávez, 2024[16]). Their skills, attitudes and engagement are key to build a culture of learning and innovation, restructure organisational workflows, and adopt AI that is in line with the objectives of their institution. Moving away from a traditional role of top-down management, leaders will increasingly have to encourage development, excellence, and autonomy among their employees (OECD, 2023[17]).
Public sector institutions need to actively manage AI use and its risks.
Chapter 2 makes a clear case for developing strong AI governance within public sector institutions, with both enablers that facilitate AI adoption, guardrails that prevent risks, and ways to ensure engagement with users. Considering the growing range of AI applications, governments need to strengthen mechanisms that protect them against risks, for example, ethics charters, audit mechanisms, and AI oversight bodies, or accountability frameworks. A particular set of risks relates to the improper use of AI by the workforce directly. Given the widespread nature of open-access genAI tools, for example, public sector institutions need to actively and strategically manage the institutional approach to AI, encompassing the individual-level use of AI by staff.
Evidence suggests that many public servants already use AI tools, most commonly general-purpose and open access genAI tools such as ChatGPT (Gillespie et al., 2025[18]). A recent survey of UK public servants in education, health, social work, and emergency services found that 22% of respondents use GenAI at work, and a majority is using it either on a daily or weekly basis (Bright et al., 2025[19]). Such numbers will likely increase over time as AI tools become more widespread. An earlier study among Canadian civil servants found an incidence of 11% of respondents reporting to have used GenAI for work (Global Government Forum, 2023[20]). While GenAI use among employees in the private sector may be somewhat more advanced, public sector leaders likely underestimate the role AI already plays in their organisations.
According to a large-scale, representative global survey, half of the respondents who reported using AI at work admitted to using it in ways that go against organisational policies, for example, by uploading sensitive information to public AI tools (Gillespie et al., 2025[18]). Within social security organisations, considering the sensitive personalised data they handle, such misuse of GenAI poses substantial risk to the individuals they serve. Some public institutions have reacted with a temporary blockage of open-access AI systems on government devices, for example, at the United States Social Security Administration (FedScoop, 2024[21]), while others have introduced guidelines or training for their employees on the use of the technology, explicitly warning that staff are not authorised to share any classified or private information with open-access AI tools such as ChatGPT. Advanced public sector organisations have responded proactively by introducing their own large language models (LLM) trained on internal content, as done by the municipality of Gladsaxe in Denmark which runs a self-hosted LLM based on GPT technology (Medaglia, Mikalef and Tangi, 2024[12]).
The evidence on workforce and skills implications of AI and current practices of AI use by staff within public administration underlines a twofold need. On the one hand, organisations increasingly require core technical and complementary skills and capabilities related to AI within the workforce. On the other hand, they need a deliberate approach to govern AI use of staff in compliance with regulations on AI and data. A deliberate approach to and active management of AI use will help avoid risks to data privacy, bias, or a lack of transparency, but it will also avoid a divide between groups of staff. Some employees may independently start adopting AI effectively and safely, while others need tailored skills development opportunities, explicit ethical guidelines, and clear incentives to start using AI tools at work.
As they build AI capability, social security institutions require a workforce that understands the benefits and risks of using AI. Leaning on the OECD Framework for Digital Talent and Skills in the Public Sector (2021[22]), and the AI Skills for Business Competency Framework by the Alan Turing Institute (2024[23]), it is possible to distinguish three key workforce groups to define skills needed for the adoption of AI systems.
General employees need to be able to interact with and use AI tools in a responsible and effective way.
Leaders need a strategic understanding of AI technology, including its potential and risks for the objectives of their organisation.
Digital and data professionals need the skills to support the development, implementation and maintenance of AI systems.
Each of these target groups have different learning needs, and training content targeted to them may be characterised as foundational, strategic, or technical. Learning needs are based on their role and tasks within an institution, and the skills they need to fulfil these tasks in relation to AI adoption (see Table 3.1).
Source: OECD with reference to the AI Skills for Business Competency Framework by the Alan Turing Institute (2024[23]).
None of the institutions interviewed for this report have analysed their skills needs for AI adoption and use in a structured way, and none have dedicated skills framework on AI in place. Yet, some are already using frameworks for digital skills that cover AI elements. The Latvian Digital Academy, for example, which is part of the School of Public Administration that provides training for almost 63 000 civil servants, developed a digital skills framework to strengthen the digital skills of government employees, in collaboration with an external provider. The framework is based on DigComp, the European reference framework for digital skills.1
The Latvian framework contains six competence areas: information and data literacy, digital communication and collaboration, digital content development, cybersecurity, digitalisation strategy and advancement, and transversal competencies. For each competence, it proposes four skill levels, from beginner to expert. One competence specifically relates to the use of AI: at initial level, employees know what AI is, understand its basic concepts and the potential applications in their job. At basic level, employees can create high-quality AI prompts, critically evaluate AI output, are aware of the impact of AI on the quality and efficiency of their work and of data security requirements. At advanced level, they are able to develop or adapt AI for specific needs and at expert level they promote the ethical and responsible use of AI, follow the development of trends of AI tools, plan their use in the institution, and form project teams that are able to use AI. The framework is a foundational tool to ensure that public sector workers have the digital and transversal skills needed to use digital technology, including AI. It is currently piloted in several regions of Latvia. Once validated, it will be used to assess skills of employees and identify individuals’ training needs.
Social security institutions, as other organisations, have three main levers to integrate new skills and build capacity for AI adoption: recruitment, outsourcing, and training. Compared to private sector organisations, hiring processes in government institutions are less flexible to respond to short-term skills and capacity gaps related to the adoption and use of AI. At the same time, recruitment may follow longer-term objectives for workforce and organisational development in a very strategic way. Job security is often high in the public sector, which translates in both an opportunity and obligation for institutions to invest in and incentivise the skills development of their staff. A challenge for many government institutions is to attract and retain good candidates, particularly in fields with skill shortages such as ICT (Eurofund, 2023[24]). To address these issues, governments have taken different approaches to create attractive, high-impact opportunities for digital and data professionals in government, and to improve the conditions and career opportunities for this occupational group.
Outsourcing is a standard alternative for government institutions, including in social security, to respond to a lack of skills or in-house capacity, and implement new AI solutions. The decision to outsource certain tasks, however, needs to be weighed against the need to stay accountable as public service providers. As mentioned in chapter 2, maintaining key skills and functions in-house and strategically managing third party providers has a range of advantages. In the context of AI adoption, there is a heightened need to guarantee data privacy and security, ensure compliance and accountability, integrate AI tools in an iterative way that supports organisational objectives, and avoid information asymmetry and vendor lock-in in procurement (Kupi, Jankin and Hammerschmid, 2022[25]; Medaglia, Mikalef and Tangi, 2024[12]). These considerations are also reflected by Finland’s national social security institution’s strategic choices in in favour of largely autonomous technological infrastructure, as mentioned in Chapter 1. One way in which public administrations manage procurement in a targeted way is through GovTech programmes, which are strategic collaborations with start-ups, public or private innovation units, or academia, to support innovative digital and AI solutions (OECD, 2024[26]).
Offering high-impact programmes to attract tech talent
Some governments have designed programmes that are targeted at digital and data professionals, typically high-skilled graduates or young professionals, and offer attractive, high-impact opportunities. An example is the Tech4Germany and Work4Germany Fellowships by the Digital Service by the German federal government which bring together experts from the private sector and federal ministries and authorities on transformation projects. The competitive fellowship places individuals in interdisciplinary teams that tackle specific projects within a few months using agile and user-centric methods. Tech4Germany focuses on product, design, and engineering professionals who are asked to create prototype software products within three months, and Work4Germany focuses on digital change management, the redesign of work processes and iterative work methods (Digital Service, 2025[27]).
Operating since 2012, the Presidential Innovation Fellows (PIF) is a similar, highly competitive fellowship programme in the United States that pairs successful technology leaders from the private sector with agencies of the federal government on high-profile transformation projects over a 12-month period. Fellows are part of an agile, multidisciplinary team with expertise in product design, software engineering or other digital and data fields. Past projects include, for example, improving data use and technological capabilities at the Center for Disease Control and Prevention, or strengthening data sharing of the Centers for Medicare & Medicaid Services within the healthcare ecosystem (U.S. General Services Administration, 2025[28]).
In France, beta.gouv.fr was established as an incubator programme that connects ambitious digital and data professionals with high-impact and time-bound digital transformation projects within French public administration. Designated a ‘start-up of the state’, it aims to attract people from the public and the private sector to different teams that each have a specific mission. Beta.gouv.fr also helps institutions setting up their own digital public service incubators and building an expert network across government. The programme follows a user-centred approach, is results-driven, and its management aims to be based on trust and autonomy for teams (République Française, 2025[29]).
These initiatives all rely on a competitive selection of high-skilled individuals with in-demand digital and data skills. They appeal to individuals with a purpose-driven mission and typically promote collaboration and teamwork among selected fellows. By placing individuals in time-bound projects with potentially large impact, they drive innovation and technology adoption within the public sector.
Improving career pathways for digital and data professionals
Social security institutions, typically through initiatives at the national level of public administration, are benefitting from various initiatives to make careers in the public sector more attractive to digital and data professionals, and to prioritise digital and data skills. This includes the definition of dedicated roles, skills frameworks and career development opportunities. While there is typically not an explicit focus on AI, these approaches are part of building workforce capability for AI adoption.
The digital skills framework developed by the Latvian Digital Academy mentioned above, for example, includes information on skills requirement for various key roles such as data analysts, ICT security managers, service developers, AI experts, cybersecurity professionals, and digital leaders, as well as for supporting roles such as project managers, policymakers, communication professionals, customer service professionals, lawyers, and human resource management specialists. There are reflections to potentially use the framework to inform the development of a new performance appraisal system and the construction of learning pathways for digital and data professionals in the future.
The Government Digital and Data Profession Capability Framework in the UK is a skills framework introduced in 2017 to improve the recruitment, retention and skills development of digital professionals in the civil service. The framework defines more than 160 skills related to core digital functions performed by government authorities. These skills can be configured into 43 job roles in six job families: architecture, data, IT operations, product and delivery, quality assurance testing, software development and user-centred design. The Central Digital and Data Office (CDDO) within the Cabinet Office manages and regularly updates the framework according to expert feedback (Gov.uk, 2025[30]; Burtscher, Piano and Welby, 2024[31]).
The Australian government has taken different steps to attract digital and data professionals and strengthen its digital capability. Since 2019, the Australian Public Service (APS) Professions has developed specific professional job families, including ‘Digital Profession’ and the ‘Data Profession’ to better define career pathways and promote mobility, training and recruitment of professionals in these fields. The Australian Public Service Commission (APSC) is responsible for further developing skills strategies, training programmes, and data-driven workforce planning (Australian Government, 2023[32]). It relies on a whole-of-country licence of SFIA (the Skills Framework for the Information Age). APS Career Pathfinder – a public-facing application – utilises SFIA to assist agencies and career seekers in understanding the roles and skill requirements for digital, data and cyber roles (Australian Government, 2025[33]). The APSC consults with agencies to identify those role and skill requirements, and this data is now also informing the procurement of digital and data tasks. The Australian government has also adopted the APS Data, Digital and Cyber Workforce Plan 2025-30, which is in line with broader government strategies and defines a co-ordinated approach to attract, develop and retain professionals in these areas, strengthen the digital capability of the public sector, and create communities of practice.
Building GovTech programmes
GovTech refers to the collaboration between the public sector and start-ups, innovative SMEs, academia and research centres to develop innovative technologies that improve public services (OECD, 2024[26]). In this model, and contrary to outsourcing, the government does not depend on large technology providers and has access to a sandbox for testing new solutions. Research on GovTech in Europe suggests that programmes may be situated at the centre of government, within government agencies, or as independent, non-profit organisations. Their role varies from facilitators to central nodes of public sector innovation. Common activities are open competitions, hackathons, accelerator programmes, pilot experiments, research and development grants, and peer learning networks (Kuziemski et al., 2022[34]).
The Mutualité Sociale Agricole (MSA), a public institution responsible for social security and health insurance of the agricultural sector in France, strategically collaborates with universities and research centers, among them the Institute of Technological Research IRT Saint Exupery, and ANITI, the Artificial and Natural Intelligence Toulouse Institute. These partnerships support collaboration with researchers on specific AI use cases and may also help to attract talent.
In Austria, the Public Procurement Promoting Innovation (PPPI) Service Centre connects public administrations and private companies. Established as a small team within the Federal Procurement Agency in 2013, it provides information and resources, a peer learning network, consulting services for government partners, and an innovation procurement platform. It also publishes open competitions on its website, inviting private firms to submit proposals for solving (AI) challenges in the public sector, for example AI solutions to support grant applications processing (IÖB, 2025[35]; Kuziemski et al., 2022[34]).
Founded in 2021, the GovTech Campus Germany is a registered non-profit organisation that focuses on connecting government institutions with technology companies to develop and scale digital solutions. The organisation’s board includes representatives of the national, federal, and regional levels of government. Its members include IT or cybersecurity firms, academic institutions, start-ups, and civil society actors. The activities are centered on creating a platform for exchange and innovation, and an ecosystem for learning (see also next section) (GovTech Campus Deutschland, 2025[36]).
GovTech collaborations also take place at the regional level. Through the 2020 ‘Catalyst project’, the Institute for Analytics and Data Science at the University of Essex partnered with the County Council in Essex and Suffolk to provide predictive big data analysis to improve services for vulnerable people. Since 2019, the University has also appointed a Chief Scientific Adviser to the Essex County Council, with the objective to improve data-driven government services (University of Essex, 2025[37]).
Training on AI in social security institutions is critical. As mentioned above, there are limitations and challenges to hiring new staff and risks to outsourcing core functions. High job stability in public administration, including in the social security sector, means that institutions must invest in the skills development of their existing workforce over time, and create an environment conducive to continuous learning. Skills development opportunities, in this context, are a strategic lever to build in-house AI capability.
The widespread availability and use of GenAI tools means that public sector organisations have strong incentives to ensure that employees have the skills and knowledge to use AI tools an effective and trustworthy way. Most if not all employees need a basic understanding of the potential, limitations, and risks of using AI tools, building on a good baseline level of digital skills.
Since February 2025, organisations that provide or deploy AI systems in the European Union, including public administrations, are legally required to ensure that their staff has a “sufficient level of AI literacy”, according to Article 4 for the EU’s AI Act.
Providers and deployers of AI systems shall take measures to ensure, to their best extent, a sufficient level of AI literacy of their staff and other persons dealing with the operation and use of AI systems on their behalf, taking into account their technical knowledge, experience, education and training and the context the AI systems are to be used in, and considering the persons or groups of persons on whom the AI systems are to be used.
AI Act, Article 4 (European Union, 2024[38])
To ensure that employees have the skills and knowledge to use AI tools in an effective and trustworthy way and to comply with the AI Act, public sector organisations are increasingly investing in AI-related training. Typically, governments offer structured training to their staff in the form of courses or workshops. Very rarely, employees have access to longer training on AI that leads to a recognised qualification or degree (see Annex Table 3.A.2). Informal learning through learning-by-doing, reading internal guidelines, participating in educational events, peer exchange and communities of practice, etc. also plays a major role within institutions and can be promoted by leadership (see section on strategic workforce development). The following sub-section provides a comparative analysis of structured training initiatives in selected OECD countries as well as a deep dive into five selected examples.
Providing training on AI for different target groups
Many social security organisations and public sector institutions more generally have started to develop and offer training on AI for their workforce. The content of AI-related training depends largely on the learning needs of the target group, which are general employees, leaders, or digital and data professionals. Training for the general population of public sector employees typically introduces AI concepts and tools, and highlights benefits and risks of the technology. An example is the short e-learning course “ABC of AI” by the Digital State Academy of Estonia, where public servants can learn about what AI is, how to use it to create better digital services, and what to consider when developing or implementing an AI project. Such types of foundational AI training can easily be scaled, and the challenge is to achieve widespread employee uptake.
Training for leaders is often focused on the strategic potential of AI use for their organisation and business area. One example is an online training course that focuses on the business value of AI by Civil Service Learning in the UK’s Government Digital Service. Training for digital and data experts may aim to strengthen specific technical skills, provide in-depth knowledge on AI regulation, and cover case studies of AI implementation. The City of Helsinki, for example, has in 2021 and 2023 provided training opportunities on data sciences and data engineering according to the need of technical staff. Across target groups, some courses are more theoretical, while others are very applied to let participants test concrete AI tools and learn from practical use cases. Some training has a product focus and is provided to support the implementation of one specific AI tool in an organisation.
In a reality where resources are constrained, there is a trade-off between the scale and intensity of AI-related training (Figure 3.1). In other words, the number of participants needs to be balanced with the duration of training. For example, a short self-paced online course requires relatively few resources to develop and deliver, and it can be scaled to a large number of participants. A multi-day, in-person course for public sector executives, on the other hand, is more expensive and can only be offered to a selected group of participants. Figure 3.1 refers to five concrete examples of AI-related training that are implemented in OECD governments. These examples are further described at the end of this section.
Schematic representation of the trade-off between training intensity (cost per participant) and scope (number of participants), by target group
Source: OECD.
Training formats range from short online courses that employees can watch in their own time, structured training programmes that take place virtually or in-person, to modular programmes that require regular participation and culminate in a certification. In Austria, for example, the Federal Academy of Administration’s School of Data Public Services offers a certificate for participants who have completed eight days of training on AI, digital, or data-related topics within three years (Verwaltungsakademie des Bundes, 2025[39]). The delivery method can be self-paced online learning content, classroom-based courses, either virtual or in-person, or interactive in-person courses and workshops.
The providers of AI-related training are usually multiple and differ across OECD countries to include national-level training institutions for the public sector, training institutions for specific sectors of government, government agencies, non-profit training providers, or technology providers. In many countries, national training institutions for the public sector, such as the School of Public Service in Canada or the Digital Government Academy in Estonia, have started to offer AI-related training. In France, the EN3S (École Nationale Supérieure de Sécurité Sociale) and the Institute 4.10 that offer some AI-related training for the French social security sector. Innovation agencies such as Digital Norway, AI Sweden, or AI4PA in Portugal provide learning opportunities across the public and private sector. In the case of GovTech Campus Germany, a registered non-profit, training on AI is provided in co-operation with private third-party providers.
The following examples highlight concrete case studies of training with different target groups, training formats, and delivery methods. Without the claim to be comprehensive, they are selected to highlight the variety of training related to AI that governments are developing and implementing. Annex A. provides a longer list of selected AI training programmes in OECD administrations.
Example 1: E-learning on GenAI by the French Caisse Nationale d’Assurance Maladie (CNAM)
Description: Introductory e-learning module to familiarise staff with the main concepts of GenAI, understand opportunities and risks, adopt responsible practices, and learn how to effectively write prompts. The e-learning is part of a larger plan to prepare the workforce for GenAI adoption and is freely available across the network of agencies that are part of the French health insurance system under CNAM. The e-learning was launched in February 2025 and has since received a 93% approval rating from participants.
Provider: Collaboration between the human resource department and AI department at CNAM, with third party support.
Target group: All employees
Delivery method: Self-paced online course
Duration: 1 hour
Strengths and challenges: Creates a standard of reference for GenAI use in line with the larger AI strategy of the institution. The format and delivery method allow to distribute the training at scale (CNAM counting around 100 000 employees). The training content is short and participation is voluntary, which raises the question of how to achieve sufficient uptake and a positive impact.
Example 2: Course on AI use in the government of Canada by the CSPS Digital Academy
Description: Introductory course on the use of GenAI in the government of Canada. Explores GenAI tools and their potential use, effective prompting, inclusive AI practices, and the detection and prevention of inaccuracies. Training delivery is iterative, and training materials are updated regularly. The goal is a wide roll-out to reach 10 000 participants in the federal public service of Canada.2
Provider: The Digital Academy of the Canada School of Public Service, the Federal learning provider, which is attached to the Treasury Board portfolio.
Target group: All civil servants in the Canadian federal government
Delivery method: Virtual classroom-based course
Duration: 2 hours
Strengths and challenges: The course allows for large classrooms with a cap at 700 participants, complements a government guide on the use of GenAI, is available for free to civil servants and helps to create a standard of reference. The training content is relatively general and only takes place at fixed dates.
Example 3: Training on Compliance with the EU AI Act by the European Institute of Public Administration
Description: In-depth course on AI compliance and risk management in line with the requirements of the EU AI Act. It aims to provide an understanding of the AI Act’s risk-based approach, its interplay with the wider regulatory environment, including the EU’s GDPR. Through interactive workshops, participants gain knowledge and methods such as risk mitigation measures for a responsible adoption and integration of AI tools within their institutions.3
Provider: The European Institute of Public Administration (EIPA), an independent training and research centre headquartered in Maastricht, Netherlands and co-funded by EU Member States as well as the European Commission through its Erasmus+ programme.
Target group: AI project managers, data protection officers, legal professionals, and other digital professionals
Delivery method: Interactive in-person course
Duration: Two days
Strengths and challenges: The intensive course allows in-depth learning on compliance with the EU’s AI Act and network building for professionals from public administrations across different EU countries. The course costs of EUR 1230 per participant and requires travel to Maastricht.
Example 4: AI Masterclass for Senior Leaders at the Irish Institute of Public Administration
Description: Training for senior managers in Irish public service to further develop their knowledge of AI and its strategic potential for policy making and service delivery. Learning goals are understanding the terminology around AI, ethical and regulatory considerations, the relevance of data governance and quality, and the potential of AI tools in different business areas. Course facilitators apply different methods including presentations, discussions, practical challenges, and peer exchange. Course costs are EUR 460 per person, to be paid by the respective institution or individual.4
Provider: The Institute of Public Administration (IPA), a body under the Department of Public Expenditure, Infrastructure, Public Service Reform and Digitalisation of the Republic of Ireland. It offers formal training as part of an undergraduate and postgraduate programme, as well as professional development for public servants across different institutions and levels of government in Ireland.
Target group: Leaders (senior public servants)
Delivery method: Interactive in-person course
Duration: One day
Strengths and challenges: The course focuses on the strategic use of AI, corresponding to leaders’ learning needs. The in-person nature of the course facilitates networking for participants. At the same time, this limits the number of participants and makes the course resource intensive.
Example 5: Training on GenAI for human-centred services by the GovTech Campus Germany
Description: Course on different AI use cases with the potential to improve the efficiency, productivity, and service quality of public administrations. Participants learn about important aspects to consider from planning to implementation of an AI tool, including on compliance, data protection, and IT security. They build a peer network and learn about strategies to overcome challenges related to the implementation of AI tools in public administrations.5
Provider: GovTech Campus Germany, a government-funded non-profit organisation, in collaboration with third-party private providers.
Target group: Digital and data professionals, other public servants.
Delivery method: Interactive in-person course
Duration: 4 hours
Strengths and challenges: The course is available at no cost for participants and fosters network building. At the same time, it is limited to 25 participants per session and only takes place in Berlin. GovTech Campus Germany’s status as a non-profit organisation enables close collaborations with the private sector but also poses resource constraints.
Governments can also rely on existing AI training and freely available learning resources, either to encourage their staff to engage with them directly, or as inspiration to develop their own course material. The City of Helsinki, for example, encourages their 39 000 employees to enrol in the open online course “Elements of AI”, in addition to offering an e-learning on AI fundamentals, monthly learning events and product-related training on Microsoft CoPilot and other genAI tools. Box 3.1 summarises free training and learning resources that are relevant for public administration.
The following examples are training materials and learning materials on AI can be relevant for government institutions, and available at no cost online.6 They are offered by research centres or private companies:
Online learning on AI in Government by AI4GOV (https://ai4gov-project.eu/home/resources/training-learning/). The EU-financed platform offers video material, event invitations, and three open online courses focused on the theme of “trusted AI for transparent public governance fostering democratic values”.
The Elements of AI interactive online course (https://www.elementsofai.com/) by MinnaLearn and the University of Helsinki. It is a massive open online course (MOOC) that takes 4-8 hours to complete and provides a certificate for learners upon completion. It provides an introduction to AI, its real-world applications, functioning, and implications for society.
The Turing Online Learning Platform (https://www.turing.ac.uk/courses) offers a wide range of online courses on responsible AI with different duration (5-40 hours) targeted either at a general and technical audience. Includes topics such as AI ethics, bias and discrimination in AI, or transparent machine learning.
Courses, events and other learning resources by Apolitical on AI in government targeted at public servants and senior government leaders (https://apolitical.co/learning-hub/government-ai-campus/) , including an ‘AI readiness check’ (https://apolitical-arc.co/). Courses last between 1-4 hours and cover topics such as the use of AI in healthcare, AI leadership, or AI fundamentals for public servants.
Companies and organisations in the private sector often face similar challenges as public institutions. Box 3.2 provides insights on training in the private sector that aims at establishing compliance with Article 4 on AI literacy of the AI Act.
The EU AI Office has compiled a repository of training initiatives in the private sector that comply with Article 4 of the AI Act, which requires providers and deployers of AI systems to ensure a sufficient level of AI literacy of its users. The following examples are relevant for public administration and social security institutions more specifically, as they provide services for the public sector or handle very sensitive data similar to social security institutions.
Gjensidige Forsikring ASA is a Norwegian insurance company that uses several AI systems. It offers different types of AI training to staff using a multi-channel approach, including an internal knowledge and communications platforms. The company developed a mandatory e-learning course on foundational AI literacy for all its employees. Specific training is also provided depending on staff roles and responsibilities, for example, training for analysts is tailored to focus on model risk management, data governance, and their specific responsibilities within the AI value chain. Staff involved in claims management is provided with product-specific training on the AI systems they use. Furthermore, when relevant, AI literacy is verified during the recruitment process, for external candidates and training on ICT and AI systems is part of the onboarding process.
Dedalus Healthcare is a global healthcare and diagnostic software company headquartered in Italy. The AI systems the company develops and uses can be classified as high-risk according to the AI Act. Training on the fundamentals of AI, its benefits and risks, ethical considerations and practical use is targeted at the entirety of the organisation's staff. Tailored training is planned for specific roles, in particular the top management, data protection officers (DPOs), developers and engineers, and legal, compliance, quality assurance and regulatory affairs departments.
Ineco is a Spanish state-owned engineering and consultancy focused on sustainable mobility and digital transformation of services in the public sector. Ineco provides company-wide AI training to all employees, focusing on foundational AI knowledge and emphasising best practices. Specialised training is also available for technical roles and project managers who are directly involved in developing or overseeing AI systems and models. This ensures that everyone has a baseline understanding of AI safety and ethics, while those working on AI projects receive more in-depth, targeted instruction. A challenge faced by the company is the shortage of qualified experts capable of designing and delivering AI training. To address this, the company provides “train the trainer” programmes for highly skilled employees, equipping them with the pedagogical tools needed to effectively teach AI-related topics to the rest of the workforce.
Smals is a Belgian NGO that provides ICT services specifically for the public and social security sector in Belgium, developing pilots, proofs of concept and prototypes for AI solutions. The organisation built a comprehensive training strategy, targeting all employees with a general AI literacy course, as well as various groups of digital and data professionals (e.g. AI experts; data engineers; infrastructure specialists; DPOs and legal teams; developers, project managers) with specific training. In addition, before initiating a new AI project, Smals conducts a product-specific training session for all relevant stakeholders. Another notable initiative in Smals is the appointment of AI ambassadors, whose role is to promote and facilitate adoption of AI technologies and maximise their value and effectiveness.
Source: European Artificial Intelligence Office, 2025, Living Repository of AI Literacy Practices v.16.04.2025, https://digital-strategy.ec.europa.eu/en/library/living-repository-foster-learning-and-exchange-ai-literacy.
Public sector organisations can strategically orient workforce development to build AI capability, including through recruitment and training, as discussed above. This section considers two elements that support AI skills and capability within the workforce, as well as the potential changes to tasks and processes within institutions, based on good practices from OECD countries: aligning workforce development with the broader AI strategy and fostering a culture of innovation and continuous learning (OECD, 2023[17]).
Aligning workforce development with the broader AI strategy
Linking workforce development to the broader AI strategy supports a more coherent approach and a better use of resources for hiring and training. In 2022, around two thirds of OECD governments reported having a learning and development strategy for their staff at the central level and at the level of ministries and agencies (OECD, 2023[17]). The link between AI and technology adoption and workforce development is not obvious, however, and organisations would benefit from taking a co-ordinated approach. Hiring and training are critical levers for the improvement of services and benefits and so is the adoption of digital and AI technology. If well aligned, these levers have the potential to augment each other.
Estonia is a good example for a strategic co-ordination of AI adoption and workforce development. It has taken a structured and gradual approach to adopting AI in the public sector through a comprehensive, whole-of-government strategy that puts an emphasis on the importance of skills development. Beginning in 2019, the country released its AI Taskforce Report7 and AI Strategy 2019-20218, marking the first steps of a carefully planned, phased rollout. This was followed by an updated AI Strategy for 2022-20239 and an AI Action Plan for 2024-2026.10 All these documents include a series of measures aimed at equipping the public sector with the skills required to effectively adopt and utilise AI.
From the outset in 2019, the AI Taskforce Report identified key workforce development measures to promote AI adoption, such as raising awareness among leadership and enhancing the skills of public officials. Subsequent strategies and action plans included specific objectives, such as the appointment of AI leaders and the provision of various learning activities on AI. More generally, the approach relies on strengthening AI and data literacy and boosting public institutions' capacity for using data.
The training and learning opportunities mentioned in the different strategies and action plan are offered through two main platforms: the Digital State Academy11 and the Kratid platform.12 The Digital State Academy is an e-learning platform for all public sector officials in Estonia developed by the Ministry of Economic Affairs and Communications in collaboration with Tallinn University of Technology. Courses are usually developed in-house by the Ministry, but some classes are designed by an external provider. It currently proposes 66 free courses covering digital and AI topics as well as other areas strategically important to the public sector, and the offer is continuously expanding. Most courses are relatively short, typically around one hour, and are intended to provide basic, introductory knowledge. Topics include data management, cybersecurity, using the government’s cloud service, AI-related courses, but also project management and equality policies for instance. The Ministry monitors course participation, completion and participants’ satisfaction. Often, the courses are used to onboard new hires.
In addition to the Digital State Academy, public officials can also access training opportunities on the Kratid platform. Managed by the Ministry of Justice and Digital Affairs, this platform presents different resources on AI and data, including an AI Support Toolbox, or training videos on data science topics. It offers detailed information on Bürokratt, the public sector virtual assistant as well as information on data governance, and the data economy. Through Kratid, the Ministry of Justice and Digital Affairs also offers several other capacity-building measures to foster AI adoption, including regular workshops and an expert network. Knowledge sharing sessions on AI use cases in the public sector are organised quarterly, featuring presentations by Estonian or international experts and discussion with the audience.
A specific AI support service is available for public institutions that want to adopt AI. The service entails brainstorming sessions and meetings to assess the needs, identify challenges, and propose potential AI solutions. Workshops tailored to the needs of specific institutions are sometimes developed, for instance an engagement session to raise awareness of AI within municipalities. Another example is a class on writing better prompts for AI tools. These specific workshops are usually developed by an external service provider.
Estonia’s approach to AI adoption in the public sector highlights the importance of integrating workforce development into a comprehensive phased national AI strategy from the outset, of offering not only general training (AI literacy) but also training tailored to the needs of different groups of professionals (leaders, AI experts), making them accessible through platforms like the Digital State Academy and Kratid, and of regularly engaging with public sector officials to ensure buy-in and wide participation in learning opportunities.
Fostering a culture of innovation and continuous learning
Previous work by the OECD has highlighted the importance of a learning culture to support a more flexible public sector (OECD, 2023[17]). This is particularly urgent in the case of the adoption of AI tools. The adoption of AI systems relies on a basic level of awareness and openness by all potential AI users within an organisation. At the same time, technological solutions may have the greatest effect if combined with a rethinking of organisational processes and tasks, which in turn require staff to reflect critically and innovatively on how to best achieve the goals of their organisation. While leaders and digital experts can certainly drive innovation on AI, the entirety of the workforce contributes to the speed and ease of change management and the effective and responsible adoption of AI. While hard to measure, an environment that encourages and stimulates curiosity, openness, and experimentation can be deliberately created, for example, through learning events or expert communities on AI. Creating and maintaining a culture of innovation and learning can be seen as preparing the fertile ground that supports improvements over time.
To support organisational learning, many public sector institutions have central innovation and technology teams. Interdisciplinary innovation teams that combine digital and data professionals with professionals that are familiar with business processes and needs can yield promising results. An example is the AI Competence Center currently under development within the Istituto Nazionale della Previdenza Sociale (INPS), the main social security institution in Italy. The Competence Centre will be based at INPS’ central directorate in Rome, and become an internal innovation, co-ordination and governance unit focusing on AI, benefitting from contributions from professionals with interdisciplinary profiles. Its objectives are to foster employee engagement and collaboration, provide technological support across the organisation when AI systems are implemented, provide skills development opportunities on AI for employees, and improve the internal communication on AI.
Another example is the AI Center of Excellence by IT Systemhaus (as mentioned in Chapter 1), which is the IT provider of the German Federal Employment Agency. The AI Center of Excellence was set up in 2019. It now consists of around 15 people, many with a data science background, but also other professional profiles. Whenever the AI Center develops a new product, they set up an interdisciplinary project team that includes both business professionals and technology experts. Over time, it became clear to them that the involvement of business professionals not only mattered to construct training sets for machine learning models but was key to develop user-centered solutions that meet business needs. Nowadays, before any AI system is implemented, data and AI ethics, data protection and IT security functions assess the risks associated with a large-scale implementation of the system, ensure compliance with GDPR and AI Act, and approve the AI system for use.
A culture of innovation and continuous learning needs to be deliberately created within organisations. An interesting example is the learning week on AI organised by the Canada School of Public Service. Civil servants in the Canadian federal government are encouraged to participate in a range of condensed learning activities during this week, including live events, instructor-led courses, and self-paced learning (Government of Canada, 2025[40]).
Another example is the yearly conference called Kratitreff in Estonia, which brings together AI experts and users from Estonia’s public and private sectors to support better decision-making and service delivery using AI, in line with the country’s broader AI strategy (see above). Similarly, the Artificial Intelligence and Data Science for Public Administration (AI4PA) initiative in Portugal offers a cycle of training events focusing on AI and digital transformation in public administration and SMEs. Day-long events encompass training sessions, interactive workshops, information on funding opportunities and IT solutions. Registration costs around EUR 30 per person and includes a certificate of completion at the end.
The Estonian Ministry of Justice and Digital Affairs runs different networks to develop an expert community where collaboration and the exchange of good practices across the public sector can take place. A data expert group gathers IT leaders from various institutions and agencies at the national level to share experiences around AI development and implementation. The Ministry also built a community of practice composed of around 700 professionals working in AI, data, or IT in public organisations such as Statistics Estonia, the national statistical agency, the public employment service, the Social Insurance Board, or the Land and Spatial Development Board. The Ministry regularly engage with the community of practice via e-mail, newsletter, conferences and workshops and sometimes ask for feedback on their needs, notably in terms of training.
Public servants, whether in general or in digital and data roles, can benefit from spaces that allow for experimentation with AI. The AI CoLab in Australia aims to provide just that, allowing public servants to learn about and test different AI tools, and connect with other institutions at different levels of government, NGOs, academia and the private sector. The AI CoLab hosts different events, workshops and training on AI, for example, a regular in-person workshop called “practical AI for policy people” that allows public servants, alongside cross sector participants, to familiarise themselves with concrete AI tools.
The City of Helsinki has a dedicated “experimentation accelerator” in place, to foster and reward innovation on AI and digital technology. The accelerator was founded in 2019 as part of the city’s digitalisation programme and is designed to support rapid internal experiments with technology and new ways of working. Over several calls for proposals, city employees handed in 120 proposals for experimentation, of which 65 experiments. Each chosen proposal received funding of around EUR 10 000 that allowed to get third-party support. Experiments do not always end with the implementation of a new technology but have the explicit objective of developing ‘lessons learnt’, which are collected on a dedicated webpage and can be built on in further projects.13
AI adoption will have an impact on the social security workforce
The successful adoption of AI systems in the social security sector has the potential to increase the productivity of services and free up staff time for higher-value tasks. Most importantly, the adoption of AI is likely to transform existing job roles and processes within organisations over time.
While AI adoption may increase the demand for data and digital professionals, a good baseline level of digital skills and AI literacy among all of staff is key. Complementary and transversal skills such as problem-solving, critical thinking, or social skills will facilitate organisational change.
The growing use of AI tools in social security organisations needs to be actively managed, including through building core skills and capabilities related to AI within the workforce by
empowering public sector employees with AI literacy;
enabling leaders to drive organisational change; and
attracting and retaining digital and data professionals.
Social security institutions need in-house AI capability
Maintaining key skills and functions in-house strategically supports higher accountability, transparency and data privacy, better alignment of AI systems with organisational objectives, and a lower dependency on technology vendors.
Governments have different tools to improve the hiring and retention of digital and data specialists, for example, high-impact programmes with timebound projects, well-defined job roles and career pathways for this occupational group, and strategic GovTech programmes.
Given the limited flexibility in hiring, the continuous skills development of staff is key. Public institutions are increasingly offering AI-related training for their employees. They may target general employees to develop foundational AI skills, leaders to gain strategic knowledge on AI, or digital and data professionals to improve technical AI-related skills.
Workforce development can strategically support AI adoption
Workforce development, including through hiring, outsourcing, and training, should be aligned with and part of the broader AI strategy of an organisation.
A strong culture of innovation and learning in public sector institutions will support change processes and the continuous improvement of services and benefits using AI, for example, through interdisciplinary teams, spaces for experimentation and incentives for continuous learning.
References
[6] Acemoglu, D. et al. (2022), “Artificial Intelligence and Jobs: Evidence from Online Vacancies”, Journal of Labor Economics, Vol. 40/S1, pp. S293-S340, https://doi.org/10.1086/718327.
[8] Atkinson, I. and J. Browne (2024), The Potential Impact of AI on the Public-Sector Workforce, Tony Blair Institute for Global Change, https://institute.global/insights/economic-prosperity/the-potential-impact-of-ai-on-the-public-sector-workforce (accessed on 12 May 2025).
[33] Australian Government (2025), APS Career Pathfinder tool, https://www.apsprofessions.gov.au/career-development/aps-career-pathfinder-tool (accessed on 24 November 2025).
[32] Australian Government (2023), Mobility opportunities | Digital Profession, https://www.digitalprofession.gov.au/career-development/mobility-opportunities (accessed on 27 June 2023).
[14] Borgonovi, F. et al. (2023), “Emerging trends in AI skill demand across 14 OECD countries”, OECD Artificial Intelligence Papers, No. 2, OECD Publishing, Paris, https://doi.org/10.1787/7c691b9a-en.
[19] Bright, J. et al. (2025), “Generative AI is already widespread in the public sector: evidence from a survey of UK public sector professionals”, Digital Government: Research and Practice, Vol. 6/1, pp. 1-13, https://doi.org/10.1145/3700140.
[3] Brynjolfsson, E., D. Li and L. Raymond (2023), “Generative AI at work”, Nber working paper series, Vol. 31161, http://www.nber.org/papers/w31161 (accessed on 12 May 2025).
[9] Bundesdruckerei and Possible (2025), Fachverfahren Decoded. Daten-und KI-Anwendungen für effiziente öffentliche Verwaltungen, https://scnem2.com/art_resource.php?sid=8vufi.j1b3di,force_download=1&u=5a59803f51917537a55471fcb5fd688f (accessed on 28 July 2025).
[31] Burtscher, M., S. Piano and B. Welby (2024), “Developing skills for digital government: A review of good practices across OECD governments”, OECD Social, Employment and Migration Working Papers, No. 303, OECD Publishing, Paris, https://doi.org/10.1787/f4dab2e9-en.
[10] Chinn, D. et al. (2020), The future is now: Closing the skills gap in Europe’s public sector, McKinsey & Company, https://www.mckinsey.com.br/~/media/McKinsey/Industries/Public%20and%20Social%20Sector/Our%20Insights/The%20future%20is%20now%20Closing%20the%20skills%20gap%20in%20Europes%20public%20sector/The-future-is-now-Closing-the-skills-gap-in-Europes-public-sector-vF.pdf (accessed on 23 May 2025).
[2] Czarnitzki, D., G. Fernández and C. Rammer (2023), “Artificial intelligence and firm-level productivity”, Journal of Economic Behavior & Organization, Vol. 211, pp. 188-205, https://doi.org/10.1016/j.jebo.2023.05.008.
[27] Digital Service (2025), Fellowships for digital pioneers and change experts, https://digitalservice.bund.de/en/fellowships (accessed on 6 June 2025).
[24] Eurofund (2023), Measures to tackle labour shortages: Lessons for future policy., Publications Office of the European Union, Luxembourg., https://www.eurofound.europa.eu/en/publications/2023/measures-tackle-labour-shortages-lessons-future-policy (accessed on 28 January 2025).
[38] European Union (2024), Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence and amending Regulations (EC) No 300/2008, (EU) No 167/2013, (EU) No 168/2013, (EU) 2018/858, (EU) 2018/1139 an, http://data.europa.eu/eli/reg/2024/1689/oj.
[21] FedScoop (2024), “More federal agencies join in temporarily blocking or banning ChatGPT |”, https://fedscoop.com/more-federal-agencies-join-in-temporarily-blocking-or-banning-chatgpt/ (accessed on 26 May 2025).
[18] Gillespie, N. et al. (2025), Trust, attitudes and use of artificial intelligence A global study 2025, University of Melbourne; KPMG, https://doi.org/10.26188/28822919.
[20] Global Government Forum (2023), Attitudes to AI in the Canada Public Service.
[7] Gmyrek, P. et al. (2025), Generative AI and jobs: A 2025 update, International Labour Organisation, https://www.ilo.org/publications/generative-ai-and-jobs-refined- (accessed on 4 June 2025).
[11] Gouvernement Français, DINUM and INSEE (2021), Évaluation des besoins de l’État en compétences et expertises en matière de donnée, https://www.numerique.gouv.fr/uploads/RAPPORT-besoins-competences-donnee.pdf (accessed on 11 February 2025).
[30] Gov.uk (2025), Government Digital and Data Profession Capability Framework, https://ddat-capability-framework.service.gov.uk/ (accessed on 27 May 2025).
[40] Government of Canada (2025), Learning Week on Artificial Intelligence, https://www.csps-efpc.gc.ca/events/learning-week-ai/index-eng.aspx (accessed on 6 June 2025).
[36] GovTech Campus Deutschland (2025), GovTech Campus Deutschland, https://govtechcampus.de/ (accessed on 27 May 2025).
[13] Green, A. (2024), “Artificial intelligence and the changing demand for skills in the labour market”, OECD Artificial Intelligence Papers, No. 14, OECD Publishing, Paris, https://doi.org/10.1787/88684e36-en.
[35] IÖB (2025), Public Procurement Promoting Innovation, https://www.ioeb.at/en/ (accessed on 27 May 2025).
[25] Kupi, M., S. Jankin and G. Hammerschmid (2022), Data science and AI in government. Why public sector organisations need in-house data science and artificial intelligence expertise, Hertie School, https://www.hertie-school.org/fileadmin/2_Research/2_Research_directory/Research_Centres/Centre_for_Digital_Governance/5_Papers/Other_papers/HS_Policy_Brief_English_Final_Version_Print.pdf (accessed on 14 May 2025).
[34] Kuziemski, M. et al. (2022), GovTech Practices in the EU, Publications Office of the European Union, Luxembourg, https://doi.org/10.2760/74735.
[4] Lane, M., M. Williams and S. Broecke (2023), “The impact of AI on the workplace: Main findings from the OECD AI surveys of employers and workers”, OECD Social, Employment and Migration Working Papers, No. 288, OECD Publishing, Paris, https://doi.org/10.1787/ea0a0fe1-en.
[12] Medaglia, R., P. Mikalef and L. Tangi (2024), Competencies and governance practices for AI in the public sector, Publications Office of the European Union, https://data.europa.eu/doi/10.2760/7895569 (accessed on 26 May 2025).
[1] Mergel, I. et al. (2023), “Implementing AI in the public sector”, Public Management Review, pp. 1-14, https://doi.org/10.1080/14719037.2023.2231950.
[26] OECD (2024), Enabling Digital Innovation in Government: The OECD GovTech Policy Framework, OECD Digital Government Studies, OECD Publishing, Paris, https://doi.org/10.1787/a51eb9b2-en.
[5] OECD (2023), OECD Employment Outlook 2023: Artificial Intelligence and the Labour Market, OECD Publishing, Paris, https://doi.org/10.1787/08785bba-en.
[17] OECD (2023), Public Employment and Management 2023: Towards a More Flexible Public Service, OECD Publishing, Paris, https://doi.org/10.1787/5b378e11-en.
[22] OECD (2021), “The OECD Framework for digital talent and skills in the public sector”, OECD Working Papers on Public Governance, No. 45, OECD Publishing, Paris, https://doi.org/10.1787/4e7c3f58-en.
[29] République Française (2025), Incubateur de services publics numériques - beta.gouv.fr, https://beta.gouv.fr/ (accessed on 29 July 2025).
[23] The Alan Turing Institute (2024), AI Skills for Business Competency Framework, https://zenodo.org/records/11092677/preview/AISkillsForBusinessCompetencyFramework_V2.pdf?include_deleted=0 (accessed on 2 June 2025).
[15] Theben, A., N. Plamenova and A. Freire (2023), “The “new currency of the future”: a review of literature on the skills needs of the workforce in times of accelerated digitalisation”, Management Review Quarterly, Vol. 75/1, pp. 495-526, https://doi.org/10.1007/S11301-023-00387-9/METRICS.
[28] U.S. General Services Administration (2025), Presidential Innovation Fellows, https://presidentialinnovationfellows.gov/ (accessed on 6 June 2025).
[37] University of Essex (2025), Working with partners in the Institute for Analytics and Data Science, https://www.essex.ac.uk/centres-and-institutes/institute-for-analytics-and-data-science/working-with-business#policy (accessed on 23 May 2025).
[16] Valle-Cruz, D., R. García-Contreras and J. Patricia Muñoz-Chávez (2024), “Leadership and Transformation in the Public Sector: An Empirical Exploration of AI Adoption and Efficiency during the Fourth Industrial Revolution”, Digital Government Research, pp. 794-806, https://doi.org/10.1145/3657054.3657146.
[39] Verwaltungsakademie des Bundes (2025), School of Data Public Services, https://www.vab.gv.at/bildungsprogramm/fachbereiche/public-management-und-governance/digital-government.html (accessed on 26 May 2025).
Note: Links were last accessed on 30 June 2025.
Note: Links were last accessed on 30 June 2025.
← 1. See https://joint-research-centre.ec.europa.eu/projects-and-activities/education-and-training/digital-transformation-education/digital-competence-framework-citizens-digcomp/digcomp-framework_en (accessed 26 May 2025).
← 2. See https://catalogue.csps-efpc.gc.ca/product?catalog=DDN321&cm_locale=en (accessed on 6th June 2025)
← 3. See https://www.eipa.eu/courses/artificial-intelligence/ (accessed on 6th June 2025).
← 4. See https://www.ipa.ie/information-systemstechnology/artificial-intelligence-masterclass-for-senior-public-service-leaders.6741.html (accessed on 6th June 2025).
← 5. See https://www.linkedin.com/search/results/all/?fetchDeterministicClustersOnly=true&heroEntityKey=urn%3Ali%3Aorganization%3A72086001&keywords=govtech%20campus%20deutschland&origin=RICH_QUERY_SUGGESTION&position=0&searchId=7e6cbf68-910f-4110-93a8-1641acbc3af1&sid=uXH&spellCorrectionEnabled=false (accessed on 6 June 2025).
← 6. These resources are mentioned for information, but are not officially endorsed by the OECD.
← 7. https://cdn-assets.inwink.com/b0269dea-af7b-4460-b29a-c40b9941c4c5/9042bd3d-7e92-45ad-bf58-28a412c3a13d (accessed 21 May 2025).
← 8. https://f98cc689-5814-47ec-86b3-db505a7c3978.filesusr.com/ugd/7df26f_27a618cb80a648c38be427194affa2f3.pdf (accessed 21 May 2025).
← 9. https://www.kratid.ee/en/_files/ugd/980182_4434a890f1e64c66b1190b0bd2665dc2.pdf (accessed 21 May 2025).
← 10. https://www.mkm.ee/sites/default/files/documents/2024-02/Tehisintellekti%20tegevuskava%202024-2026.pdf (in Estonian, accessed 21 May 2025).
← 13. See https://kokeilukiihdyttamo.hel.fi/results (accessed 18 June 2025).