This chapter provides lightweight coverage of how Slovenia leverages artificial intelligence (AI) to enhance government productivity, responsiveness, and accountability. It identifies opportunities to help Slovenia enhance its AI maturity to better meet its goals and serve its citizens and residents. Particular attention is paid to the key enablers, guardrails, and engagement approaches that are important for trustworthy AI adoption in government.
Digital Government Scan of Slovenia
Abstract
Artificial intelligence (AI) is one of the most transformative forces of the 21st century and is becoming a significant component of the digital government journey. Overall, the use of AI in government can improve government productivity (efficiency and effectiveness), responsiveness and accountability.
Yet, unlocking AI’s benefits requires mitigating its risks. These include ethical risks that can drive adverse outcomes and rights infringements, operational risks cause errors and erode trust, exclusion risks that widen digital divides, and public resistance to government AI. Governments’ failure to leverage AI also represents risk, resulting in missed opportunities.
Besides navigating risks, governments also face challenges when it comes to putting AI into practice. They face unique contexts and challenges that hinder the rapid uptake of AI, including skills shortages, outdated legacy systems, inadequate data and a financially constrained environment, as well a duty for public good that goes above that of companies.
The OECD Framework for Trustworthy AI in Government helps governments seize AI benefits, mitigate risks, and overcome implementation challenges (Figure 7.1). It is organised around three pillars: 1) enablers to facilitate adoption, guardrails to guide the trustworthy use of AI, and 3) engagement approaches to shape user-centred and responsive adoption.
Source: (OECD, 2025[39]).
Like many countries, Slovenia is increasingly focussed on how to leverage AI to improve its government operations and better serve its citizens. This includes adapting and building on important digital government efforts discussed in other chapters of this scan (e.g., cloud, data and data infrastructure), as well as devising initiatives and approaches specific to AI in government. Overall, Slovenia is taking significant positive steps to adopt effective and trustworthy AI in government, especially over the last year. Though as discussed in the later section on “remaining challenges”, additional efforts may be needed to ensure follow-through and implementation. According to the 2025 OECD Digital Government Index (DGI), Slovenia scores 0.63 out of 1.00 in government AI maturity, just below the OECD average of 0.64 (Figure 7.2). In comparison to the 2023 DGI, where Slovenia achieved a score of 0.57, its performance increased by 0.06 points, representing a 10% improvement. By contrast, the OECD average rose by 0.10 points, or 18%, from its 2023 score of 0.54. These figures indicate that Slovenia’s AI maturity, as measured by the DGI, advanced at a slower rate than the OECD average.
The government AI-maturity component of the DGI assesses how well central or federal governments are prepared to use AI strategically and responsibly in the public sector. It measures the presence and development of key enablers, such as national public-sector AI strategies, the inclusiveness of their design (such as stakeholder collaboration and public consultation), the existence of certain guardrails such as oversight bodies and monitoring, and the implementation of transparency instruments for algorithm use. It also evaluates the extent to which AI is used to improve government operations, policymaking, and public services.
Source: (OECD, forthcoming[40]).
Use of AI in government
Governments are increasingly using AI in different areas (Figure 7.3). The 2025 Digital Government Index (DGI), which considered initiatives in place in 2023-2024, showed that Slovenia had implemented AI for internal processes, but not for the other categories. With regard to internal processes, Slovenia’s efforts include:
Using machine learning (ML) techniques in Pladenj/Tray, its central data exchange platform, to enhance its stability and mitigate disruptions (Interoperable Europe, 2024[41]).
Employing automated transcription is improving access to court proceedings. This technology rapidly converts spoken words into written text, ensuring greater transparency and allowing more people to review and understand legal proceedings (OECD, 2025[39]).
Experimenting with AI-assisted reading and classification or filings and drafting of responses for customer service agents in the tax administration (Slovenia FURS, 2025[42]).
Note: “Oversight and accountability in the public sector” was not measured in 2023, so no comparison is possible.
Source: (OECD, forthcoming[40]).
Although Slovenia’s 2021-2026 AI strategy (Box 7.1) indicates pilots in other areas (e.g., public procurement monitoring, anti-fraud analytics), there was little evidence available to indicate these were either active, or had been completed with lessons learned derived from them. As a result, efforts not related to internal processes were not counted in the DGI.
Generally with primary implementation outside the time range of the DGI, this scan has uncovered addition efforts to experiment with more types of AI. For instance, non-exhaustive examples include:
Public service design and delivery: The country is integrating AI into national traffic management through several initiatives. The National Traffic Management Centre consolidates road and traffic data for improved mobility insights, and the DARS AI Initiative leverages ML to predict congestion and accidents on motorways (OECD, 2025[43]). Separately, the tax authority is rolling out virtual assistants to assist taxpayers in a variety of areas (Slovenia FURS, 2025[42]). 1
Policymaking: The government’s Business Intelligence (BI) platform Skrinja/Chest includes a central data warehouse that aggregates cross-government procurement data (publicly available online), data on administrative procedures, inspection procedures, common agricultural data, and data on public servants' salaries to support service-delivery review and data driven decision making using advanced analytics. However, the pilot project using ML for predictive analytics is under development and will be in place in 2026. A government research project is developing an AI model to design better policies for attracting highly educated Slovenians back to the country, providing a data-driven basis for future “brain-gain” strategies (ASEF Institute, 2025[44]).
Oversight and accountability: The financial administration uses AI in processing value-added tax (VAT) returns to increase the effectiveness of controls through predictive analytics (OECD, 2025[43]). While not a government project per se, the Slovenian Commission for the Prevention of Corruption (CPC) collaborated in an academic project to apply ML to 17 years of Slovenian public spending and gift disclosure data to build an interactive framework that helps the CPC flag suspicious companies and gifting patterns for further scrutiny (Joksimović, 2023[45]).
Slovenia has made progress in adopting AI in government, but on average, generally remains somewhat behind other European Union (EU) member states (Slovenia MKKR, 2025[46]).
Putting in place elements of the OECD Framework for Trustworthy AI in Government
Overall, Slovenia has taken important steps forward in enabling trustworthy and effective AI adoption in government, as discussed in this section. Opportunities exist to address remaining gaps and weaknesses, however, as discussed in the section on “remaining challenges”.
Enablers to facilitate AI adoption
Enablers are the foundational elements and resources necessary for AI implementation in government. In addition to the topics discussed below, digital government investment capacity, data, and digital public infrastructure (DPI) also represent critical enablers. Their importance is broader than AI alone. As such, most aspects of these enablers are discussed in corresponding chapters of this scan.
When it comes to strategy and governance, as discussed in Chapter 3, Slovenia has a national digital government strategy in place. Like most other countries, Slovenia does not have a dedicated strategy for AI in government, but instead has such an instrument embedded within a broader national AI strategy (Box 7.1). It was developed in collaboration with public sector institutions, businesses, civil society and academia, and it benefited from stakeholder workshops and an open public consultation. An inter-ministerial working group is charged with coordinating its implementation (OECD, 2025[43]). Given that the strategy is nearing its sunset date, the timing of this scan represents a good opportunity to help Slovenia take stock of progress and surface gaps to inform the design and implementation of a future strategy.
The “National Programme to Promote the Development and Use of AI in the Republic of Slovenia by 2025” (extended to June 2026) explicitly recognises government’s role as a “direct user and co-creator” of AI, with the public administration expected to procure and co-develop tailored AI solutions.
The strategy cites lessons learned from successful AI pilots in government as a key strength, but a lack of awareness of AI actors across sectors, a lack of coordinated efforts to promote AI development, and inadequate capacities, including skills, to integrate AI in the public sector as weaknesses.
The importance of AI in government is woven through many of the strategy’s “strategic objectives” (SOs), with relevant provisions listed below. Many efforts have an indicative budget figure in EUR for planned investments (generally a split between EU and budgetary funds), as indicated in parentheses.
SO1: Setting up a dynamic ecosystem. Central and inter-ministerial coordination of AI adoption and standards-setting as relevant to both AI in government and beyond. Development of at least one Digital Innovation Hub that serves both the private and public sectors.
S02: Education and strengthening of human resources. Public servant AI re-skilling programmes (500 000).
SO3: Supporting AI research and innovation. Support for a consortia of interdisciplinary innovation projects, including in public administration.
SO4: Deploying AI in key sectors. Capacity building through AI seminars for senior public managers (250 000), training of public servants on ethical and legal issues (250 000), and support for young AI researchers in the public sector. Training on specific AI models and methods targeting development teams from companies and businesses (500 000). Using AI to enhance existing digital government portals. (e.g., GOV.si). Deployment of government AI projects (23 million). Leveraging existing and new AI projects to develop and deploy fundamental building blocks for re-usable AI solutions in government.
SO5: Establishing technical infrastructure. Positions OPSI (the national open government data portal) as a foundation for data re-use for public sector AI. Analysis of the mechanism and definition of the legal and ethical framework for managing non-personal data. Creation of a national data space to enable AI in the public administration, including digital environments for pilots, prototypes, and deployment (2 million). Creation of a national platform of Slovenian AI tools (800 000). Hackathons for AI solutions that use government data (80 000).
SO6: Strengthening security with AI. Pilot projects on using AI for cybersecurity in the public administration (1 million) and promoting the trustworthy use of AI in policing (500 000).
SO7: Increasing public confidence in AI. Indicates activities will be planned and coordinated centrally, with domain-specific implementation in ministries. Each ministry should have a person responsible for AI. Establishes the position of “AI Ambassador” to communicate on AI activities and build trust. Development of a platform for sharing good practices and case studies on AI deployment across sectors, including public administration. Awareness-raising activities among companies, public sector organisations, and the public on AI risks and opportunities (250 000).
SO8: Appropriate level and ethical framework. Analysis and development of a legal and ethical framework for AI, including in public sector decision-making that affects individuals, alongside raising awareness on these issues. Establishing a framework and national oversight mechanism for certifying and monitoring AI-based solutions in alignment with the European Union’s framework for trustworthy AI.
SO9: Strengthening international cooperation. Active engagement bi-laterally with like-minded countries and in relevant international fora, including the OECD and its Global Partnership on AI (GPAI).
S010: Establishing a national AI observatory. A portal and indicator framework that tracks AI development, deployment and potential across all stakeholder groups, including public administration, to support continuous planning, monitoring and adjustment of government AI policy and investments (370 000).
Across the 10 SOs, the strategy assigns approximately 69 measures and 86 indicators to track them.
Source: (Government of Slovenia, 2021[47]).
OECD (2022[48]) work has found that a strong strategic approach for AI in government—such as in the form of an strategy, action plan, or both—includes 1) objectives and specific actions, 2) measurable goals, 3) responsible actors, 4) time frames, 5) funding mechanisms, and 6) monitoring instruments. Overall, the strategy is strong on all aspects, which is not a common achievement in such instruments. In drafting its next strategy and action plan (discussed below), Slovenia is encouraged to continue focusing on these elements, with some opportunity to make the measurable goals more outcome-oriented, and time frames somewhat more granular.
Slovenia is in the process of develop a new strategy, and it published the draft National AI Strategy 2030 (“2030 strategy”) for public consultation in November 2025. 2 Like the current strategy, it includes a solid focus on AI in government. Instead of having strategic-level goals and implementing actions in the same document, it includes 10 strategic principles, with concrete activities moved into separate forthcoming action plans. This would address a weakness identified in a recent government-commissioned evaluation of the current AI strategy (Slovenia MKKR, 2025[46]), which recommended decoupling high-level strategic objectives from granular activities to promote a more agile approach. The draft 2030 strategy’s goals include accelerating AI adoption in government, enhanced public services and public sector efficiency, as well as building institutional capacities for strategic steering, risk assessment, and impact monitoring of AI. An explicit overarching aim is to help remedy persistent challenge the OECD (2025[39]) has observed among most countries: moving from scattered pilots toward systematic, responsible embedding of AI into public administration and public services. The draft strategy includes several horizontal pillars relevant for AI in government:
National data and compute infrastructure covering supercomputing, cloud, data infrastructure, and experimentation.
Data spaces and a national framework for managing them, strengthening data stewards in government organisations, and better provision of AI-suitable government data.
Software infrastructure, including domain-specific models adapted to the Slovenian language, legal context, and cultural norms.
Competencies and talent, including training of at least 500 civil servants per year on AI governance, ethics, security and use of AI tools.
Ethics and regulation, including AI Act implementation across sectors; a National Council for AI Ethics (discussed below); a system for Fundamental Rights Impact Assessments (FRIA) for high-risk systems, with a target of 100% compliance for high-risk government systems; and AI Trust Hubs to support companies and government agencies with compliance, audits and risk management.
It also aims to put in place a permanent AI forum to provide advice to government and an AI observatory to monitor strategy implementation. Although the more-granular action plans are not yet developed, the draft has the hallmarks of a strong strategy. Yet, implementation will be the most important factor. Many of the commitments of the draft 2030 strategy are similar or the same as those in the existing strategy, in part due to incomplete implementation and challenges in monitoring the 2021-2026 strategy objectives (Slovenia MKKR, 2025[46]), as discussed in the section on “remaining challenge”.
Overall, official coordination of the strategy and government AI activities aligns with international best practices. It charges a central ministry – originally the Ministry of Public Administration, with tasks transferred to a newly created Ministry of Digital Transformation (MDT) in 2022 – with leading coordination and monitoring implementation, with line ministries responsible for implementing measures in their sectors. The ministries are brought together through an inter-ministerial coordination group. The draft 2030 strategy is envisioned to do the same.
Slovenia is working toward creating spaces to experiment with AI. Until recently, no laws or policies sought to promote experimentation in the public sector, but Slovenia’s new law on the implementation of the EU AI Act puts in place not-yet-operational provisions and governance for AI experimentation sandboxes.3 The draft 2030 AI strategy also emphasises experimentation. The Smart digital public services strategy (2024-2029) also commits to approaches that would provide the technical foundations for AI experimentation through establishing an interoperable ecosystem supporting algorithmic and analytical tools (OECD, 2025[43]).
With regards to fostering skills and talent, Slovenia has put in place several upskilling initiatives for public servants. Its Strengthening the Digital Skills of Civil Servants programme aims to improve competencies in core digital skills, including specialised skills on AI (OECD, 2025[43]). It was designed in accordance with the OECD (2021[12]) Framework for Digital Talents and Skills in the Public Sector. Under this initiative, the country’s Administrative Academy (Upravna akademija) offers a course on practical use of ChatGPT, which is one of the most in-demand courses among public servants (GOV.SI, 2025[49]). An advanced-level course also available.4 The Academy also offers a course on ethical challenges and responsible use of AI.5 These recent efforts represent very positive strides forward. Slovenia does not have its own competency framework to standardise AI skills across the public administration, but the EC Joint Research Centre (JRC) has a comprehensive framework and can be leveraged by any EU member state.6 Slovenia has typically leveraged broader competency frameworks like this for digital skills.7
Establishing guardrails to guide trustworthy AI
As an EU member state, Slovenia is bound by the requirements of the EU AI Act and GDPR, which includes many safeguards, and transparency and accountability measures relevant to AI (especially for systems designated “high-risk”) and data protection more broadly. Slovenia also employs non-binding levers; the national AI strategy integrates the OECD AI Principles and EU ethical AI guidelines and emphasises ethical principles throughout the programme, such as to assure proper regulation, promote trust, and understand impacts through open and expert consultations. Slovenia adopted a law to implement the EU AI Act, which went into force in November 2025.
With regard to advancing accountability and risk management, the draft 2030 strategy roots Slovenia’s risk management practices in recognised framework (e.g. the Council of Europe’s HUDERIA methodology and the United States NIST AI Risk Management Framework), presenting them as references for a national approach, though this document is a draft and the how-to of the commitments is yet to be determined. The Information Commissioner has also issued guidance on how to protect personal data when developing and using AI systems.8 When it comes to the use of generative AI tools in government in particular, Slovenia has issued guidance to public servants, touching on data protection and security rules, human oversight and professional responsibility, transparency, education and training, and respect for law, ethics, and intellectual property. 9
With regard to oversight and advisory bodies to help oversee and guide trustworthy AI adoption in government, data for the 2025 DGI show that Slovenia does not have a central body in charge of providing overseeing AI in the public sector. Instead, it has put in place a distributed system across several authorities, with MDT as a strategic lead, the Information Commissioner as the main fundamental rights and data protection supervisor, and the Agency for Communication Networks and Services of the Republic of Slovenia (AKOS) as a key market monitoring authority. Additional AI-related responsibilities for these actors are articulated in Slovenia’s new law on the implementation of the EU AI Act, such as charging AKOS with implementing AI experimentation sandboxes.10 With regard to advisory bodies, the government has launched a public call for nominations to appoint members to a National Council for Ethics in AI, an autonomous and consultative body tasked with advising on ethical, human-rights, and broader societal issues related to AI, including its responsible use in the public sector (2025[50]) (GOV.SI, 2026[51]). The draft 2030 strategy further proposes a permanent AI Forum and a national AI competence centre to provide strategic, multi-stakeholder advice and support across the economy and the public sector.
Engagement to shape strategic and responsible AI
As discussed above, both Slovenia’s current and draft future nation AI strategy was developed through significant engagement and collaboration with relevant stakeholders inside and outside government. The law for implementing the AI Act also benefited from open consultation.
Overall, Slovenia has put forward solid and well-considered strategies and commitments for AI in government. Its ambitions are clear and aligned with OECD frameworks and international best practices. However, it has struggled at times to achieve strong coordination, follow-through and implementation of its objectives. Opportunities exist for Slovenia to take targeted action to help ensure successful implementation of its forthcoming 2030 AI strategy and associated action plans.
Challenges instilling key enablers
While the current strategy has all the hallmarks of a strong strategy, there are some weaknesses in its design and implementation. While strategies demonstrate leadership and establish a vision and commitments, subsequent implementation action is critical to achieving the strategic goals and objectives. Although the strategy defines solid indicators and envisions a strong approach to monitoring, evidence of actual monitoring and progress tracking is light and generally not public.
A 2025 (2025[46]) study commissioned by the Ministry of Cohesion and Regional Development (MKRR) found mixed results with regard to strategy (Box 7.1) implementation. Only SO1 was fully met, with all other objectives partially achieved. Particularly relevant to AI in government, SO4 (deploying AI in key sectors) achieved 1 of 18 indicators and SO5 (infrastructure) achieved 3 of 11. However, the study states that actual progress in deploying AI in key sectors, including public administration, could be greater than the indicator scores suggest, because of lack of coordination, absence of a unified monitoring system, fragmented responsibilities, fragmented indicators and data sources. While coordination within certain ministries – such as the Ministry of Defence’s Working Group on AI – coordination across ministries is limited (Savič, 2024[52]). No visible central AI leader appointed (e.g., a Chief AI Officer) and AI leads within ministries have not been determined, and the interministerial co-ordination arrangements foreseen in the draft 2030 strategy and AI Act implementation law remain at an early, mainly procedural stage, which may be insufficient on their own to secure coherent follow-through on AI strategy commitments. Unless Slovenia strengthens these areas, it may have similar difficulties implementing and tracking progress in achieving its forthcoming 2030 strategy.
Bolstering the skills and capacities of public servants is a critical area needing improvement. The MKKR (2025[46]) study found AI deployment in the public sector is fragmented, with varying maturity, little knowledge transfer and no common directions or approaches. Focusing on upskilling existing civil servants, rather than focusing on recruitment of skilled individuals not already in the public service is perhaps particularly important in Slovenia. Overall, the country has one of the lower AI skills penetration rates in the EU (Figure 7.4), and net AI skills migration into the country has been negative, which may limit recruitment opportunities. While Slovenia has a variety of valuable initiatives on digital skills in government, only a few initiatives exist specific to AI. As discussed above, courses exist on practical use of AI tools and ethical use and AI ethics, but data from the 2025 DGI and subsequent follow-up research show nothing on data privacy and security, or implementation in public services or policymaking, as specific to AI. A recent focus group of 26 Slovenian public sector participations found they had significantly different understandings of AI, with many lacking visibility of any concrete, practical examples in the Slovenian public sector (Slovenia MKKR, 2025[46]). Key barriers identified from the focus group included a lack of knowledge and competencies, especially targeted training for advanced uses and specialisations; as well as a lack of knowledge transfer and any central space for sharing good practices and coordinating AI efforts between government organisations, leading to duplication of effort.
Note: This chart shows the prevalence of workers with AI skills – as self-reported by LinkedIn members from 2016-2024 – by country and against a global average benchmark. An AI skills penetration of 1.5 means that workers in that country are 1.5 times more likely to report AI skills than workers in the benchmark. Please see (OECD.AI, 2025[53]) for more information.
Source: (OECD.AI, 2025[54]).
Slovenia has few opportunities for informal learning, which in addition to training could assist in addressing these challenges. Such activities could facilitate connections and knowledge exchange in the form of communities of practice or networks. Conferences such as the “Tomorrow – Public Administration” (Konferenca JUTRI – Javna uprava) and the Days of Slovenian Informatics provide an annual chance to connect and share on digital and other topics, and some local-level knowledge sharing exist, but opportunities exist for a more ongoing community of central government civil servants working on, or otherwise interested in, AI and other digital topics to convene and collaborate.
Although they are more specifically grounded in other chapters of this scan, data from the 2025 DGI data show that there is no specific investment decision-making criteria for public sector AI, and that the central government is not supporting public institutions in procurement of AI. Focus group participants also listed infrastructure limitations as a barrier. There is also no dedicated funding instrument that finances AI deployment (Slovenia MKKR, 2025[46]). This practice is not always needed, but it is worth considering helping achieve a consolidated and strategic view of the government’s AI portfolio. Although the current strategy includes envisioned investments, these do not represent guaranteed resources. Although government officials report allocating around EUR 110 million by 2025 to implement the broad strategy, funding for AI in government is scattered with little strategic direction, which as also emphasised in the MKRR (2025[46]) study.
Guardrails
As touched on above, the requirements of the EU AI Act and GDPR address many aspects of putting in place guardrails for trustworthy AI. Yet, there are still Slovenia-specific actions that could be taken to further ensure the effective and responsible use of AI in government.
Data from the 2025 DGI show that Slovenia has few internal controls or mechanisms in place to ensure accountability in the development and deployment of AI systems in the public sector. In particular, it lacks algorithmic auditing processes, internal review committees, ex-ante (i.e., before deployment) risk assessment requirements and tools, citizen complaint mechanisms, and data origin tracking. The EU AI Act and related Slovenian law to implement it require some risk management activities for high-risk systems, and the draft 2030 strategy envisions some of these instruments, though present-day activities are limited. GDPR also includes requirements for ex-ante assessments and citizen complaint mechanisms for systems using personal data. Slovenia could further ensure its use of AI is trustworthy by prioritising establishment of required protocols and those already described in the draft 2030 strategy.
When it comes to the use of generative AI tools in government in particular, although Slovenia has internal guidance on ethical guidelines for using GenAI available to public servants, the country generally lacks some important measures, including:
Regular audits to ensure compliance with AI policies and regulations.
Independent oversight by external bodies or experts.
Transparency measures, such as disclosure of AI-generated content.
Accountability frameworks to address potential misuse or errors.
The EU AI Act and Slovenia’s law to implement it touches on some of these. This includes independent oversight through the Information Commissioner and National Council for Ethics in AI, requirements disclosing to users when they are interacting with an AI system, and accountability requirements when deploying high-risk GenAI systems.
In practice, as of the end of 2024, data from the 2025 DGI indicate that Slovenia’s impact measurement activities are limited. It had not conducted any financial or non-financial impact measurement studies on AI in government as related to a specific AI use case, for the impact of AI across a specific government sector, or for across the government more broadly. Some limited efforts have looked at the impact of government AI use on the broader economy and society. This makes it challenging to identify potential or actual impacts of government use of AI systems within and beyond government, limiting opportunities to mitigate risks and negative impacts, and realise benefits and other positive impacts.
In terms of promoting transparency, Slovenia has not yet developed a transparent registry of AI use cases in government, which has increasingly become a best practice (OECD, 2025[39]). It may submit some cases to the EC Public Sector Tech Watch observatory,11 which shows use cases from across the EU identified by EC desk research or submitted directly from countries. As of July 2025, the Tech Watch data lists 11 Slovenian AI use cases across levels and functions of government.12 A civil society organisation called Danes je nov dan has created one, listening nine AI systems and suggesting that dozens of others are also in use (2025[55]). 13 Slovenia’s AI Act implementation law requires creation of a central public information point where public sector AI systems will be listed,14 and mandates the responsible ministry to promote public awareness of those systems, though other countries have benefited from a more comprehensive registry. MDT has indicated it will explore the development of such a registry (RTV SLO, 2025[56]; Danes je nov dan, 2025[57]).
Engagement
Overall, while engagement activities around core products (e.g., national AI strategies and laws) has been commendable, engagement around the use of AI in Slovenia’s public sector remains ad hoc. More could be done to build systemic, participatory engagement with users, citizens, and social partners on government use of AI. Such efforts seem more prevalent in other digital government areas, such as open government data, where MDT has, for example, used International Open Data Day and related OPSI workshops to bring together data publishers and users, and has commissioned surveys to understand how open data are used in practice. Some of the most innovative engagement pressure is currently coming from civil society, such as Danes je nov dan’s public sector AI registry (an independent source not verified by government), rather than from formal government mechanisms.
Based on these findings, the government of Slovenia could consider the following recommendations in the finalisation and implementation of its new 2030 AI strategy, forthcoming action plans, and other related initiatives.
As it seeks to expand its integration into the public sector, Slovenia could aim for a more implementation-oriented and coordinated approach to AI, promoting AI adoption aligned with the OECD Framework for Trustworthy AI in Government.
Recommendation 17:
Strengthen coordination governance and leadership. Prioritise the establishment of the permanent interministerial co-ordination group foreseen in draft 2030 strategy and ask key ministries and major public bodies to nominate knowledgeable and accountable AI leads so that strategic priorities, implementation, and lessons from pilots are steered coherently across government, and that clear lines of accountability exist for ensuring progress in implementing strategic objectives and action plans. Consider the establishment of a central Chief AI Officer (CAIO) or similar position to chair the group and champion AI in government efforts across the public administration.
Recommendation 18:
Strengthen implementation monitoring and portfolio oversight. Establish a clear implementation monitoring framework for AI in government by tasking the planned AI observatory with tracking a consolidated portfolio of AI systems in government, using a small set of agreed indicators on uptake, risk level and impacts, and reporting regularly to the interministerial co-ordination group, the AI Forum and political leaders so that strategy commitments translate into concrete, measurable progress.
Recommendation 19:
Develop practical, enabling guidance. Develop cross-cutting and domain-specific (e.g., procurement of AI) guidance that helps public servants understand how to design, buy and operate AI systems in practice, not only what to avoid, for example step-by-step playbooks, model documents and checklists for AI-related procurement, risk assessment, contracting and lifecycle management, with concrete examples and templates that ministries can readily adapt.
Recommendation 20:
AI skills and a cross-government community of practice. Build on recent efforts to promote AI-related skills among civil servants by further investing in training that covers currently under-served topics such as AI in public services or AI project delivery, procurement, evaluation and risk management; and by establishing a cross-government community of practice on AI that brings together practitioners from ministries, agencies and local government to share lessons, examples and practical solutions.
Recommendation 21:
Accelerate AI adoption through targeted support by the Competence Centre for AI. Ensure that AI adoption in government is a core, explicitly mandated focus of the envisioned AI Competence Centre by creating a dedicated government workstream with tailored advisory services, training and project support for ministries, and by prioritising high-impact use cases and shared solutions that can be scaled across the public sector.
Recommendation 22:
Consider an AI-specific funding mechanism for AI innovation. Consider creating a dedicated AI funding mechanism to back high-priority use cases and shared enablers, linked to clear criteria on public value, risk management and learning.
Recommendation 23:
Develop proportionate pre- and post-deployment impact assessment for AI in government. Design a standard, proportionate approach for ex-ante and ex-post impact assessment of AI systems in government, including simple templates and guidance that help teams identify potential and actual effects on service quality, costs, human rights and equity, and require high-risk systems to undergo more detailed assessment, documentation and follow-up learning.
Recommendation 24:
Develop a transparent public registry of AI systems in government. Establish and maintain a comprehensive public registry of AI systems used across government that discloses basic information on each system’s purpose, such as provider, data use, risk level and human oversight; and integrate it into routine project approval and reporting processes so that new and modified systems are systematically recorded, searchable and open.
← 2. https://www.gov.si/zbirke/javne-objave/javna-obravnava-predloga-nacionalnega-programa-za-umetno-inteligenco-2030-npui-2030.
← 3. Act on the Implementation of the EU AI Act laying down harmonised rules on AI (Zakon o izvajanju uredbe (EU) o določitvi harmoniziranih pravil o umetni inteligenci – ZIUDHPUI. See https://pisrs.si/pregledPredpisa?id=ZAKO9225 for the official law, and https://www.dataguidance.com/opinion/slovenia-draft-act-implementation-eu-ai-act for an English summary.
← 4. See https://ua.gov.si/aktivnosti/detajli/?ID=6c36c932-a6fd-ef11-aba1-005056817c25&Tag=459,576 (basic course) and https://ua.gov.si/aktivnosti/detajli/?ID=ee6c77d1-d39a-f011-abb2-005056817c25&Tag=576,459 (advanced course).
← 7. For instance, Slovenia typically uses the Digital Competence Framework for Citizens (DigiComp 2.2) to assess the digital skills of public servants. See https://publications.jrc.ec.europa.eu/repository/handle/JRC128415 for more information.
← 8. https://www.ip-rs.si/zakonodaja/reforma-evropskega-zakonodajnega-okvira-za-varstvo-osebnih-podatkov/kljucna-podrocja-uredbe/umetna-inteligenca-in-varstvo-osebnih-podatkov.
← 9. https://nio.gov.si/en/products/priporocila%2Bjavnim%2Busluzbencem%2Bpri%2Buporabi%2Borodij%2Bgenerativne%2Bumetne%2Binteligence%2Bdostopnih%2Bna%2Bspletu.
← 10. See https://e-uprava.gov.si/si/drzava-in-druzba/e-demokracija/predlogi-predpisov/predlog-predpisa.html?id=17671&lang=si for the bill, and https://www.dataguidance.com/opinion/slovenia-draft-act-implementation-eu-ai-act for an English summary.
← 14. This obligation does not apply to AI systems, or AI-enabled information solutions, used for the prevention, detection, or investigation of criminal offences, or for defence, national security, migration, asylum, or border management. It also does not apply to systems used exclusively for military, defence, or national security purposes. The legislation further establishes oversight of compliance with the public disclosure requirements for AI systems. The Information Society Inspectorate responsible for supervision and enforcement.