50 institutions in Croatia use or test artificial intelligence (AI) in their work, according to Gong’s research, "AI in the Public Sector: How Croatia Uses Artificial Intelligence", conducted among 78 key public authorities. Of the 50 institutions that officially use or test AI, as many as 26 said they have no internal rules governing its use.
We conducted the research by sending requests for access to information to institutions through Gong’s digital platform ImamoPravoZnati.org, asking them which AI solutions they use or are testing, whether they have adopted specific rules governing the use of AI, and whether they have rules for their employees on the use of generative AI systems.
We identified 55 different AI solutions currently being used, across areas ranging from healthcare, justice and security to administration, finance and communication with citizens.
From Generative to Highly Specialized AI
Institutions most commonly use the generative AI tools Microsoft Copilot and ChatGPT, as well as ANON, a system for anonymizing court decisions. Other solutions identified in use include Siemens syngo.via and the Da Vinci Surgical System for diagnostics and surgery, ABIS and NABIS for automated biometric identification, and Face Recognition for facial recognition. We also identified a number of AI-powered virtual assistants, such as HANA, GINA and ZdrAVKO in healthcare, the GlasAI virtual assistant developed by the State Election Commission, and others.
The largest number of different AI solutions, 31, are used for specific professional tasks within institutions. By function and area of application, three are used for analytics and prediction, 15 for processing information and documents, eight for healthcare and medicine, three for biometric identification, and two for cybersecurity.
You can find out which public authorities use AI, how they use it, and access all other data in the interactive table at this link and in the research report and its appendices at this link.
AI is already present in areas of the public sector that directly affect the human rights and freedoms of citizens. Yet Croatia still lacks a comprehensive overview of how AI is used by public authorities, as well as consistent rules governing its use. Publicly available data from Croatian institutions, the European Commission and the OECD provide only a fragmented and limited picture of AI use in Croatia’s public sector. This prompted Gong to conduct the first more systematic mapping of AI use in Croatia’s public sector between February and August 2026.
Government of the Republic of Croatia: We Do Not Use AI
26 public authorities told us that they do not use AI, including the Government of the Republic of Croatia and 10 ministries (of Defence, Finance, Foreign and European Affairs, Science, Education and Youth, and others). Other public authorities that reported not using AI include the Constitutional Court, the State Attorney’s Office, the Office for the Suppression of Corruption and Organized Crime (USKOK), the Tax Administration, the City of Zadar, two counties, and other independent bodies and agencies.
Nine institutions told us that they use AI and have specific rules governing its use, including the Croatian Parliament, FINA, the City of Zagreb, APIS IT, HAKOM, HANFA and the Croatian Institute of Public Health (HZJZ).
A further 10 institutions told us that they use or test AI and that, at the time of the research, they were preparing specific rules governing the use of AI. These included three ministries (the Ministry of Justice, Public Administration and Digital Transformation; the Ministry of Physical Planning, Construction and State Assets; and the Ministry of Labour, Pension System, Family and Social Policy), the Croatian Personal Data Protection Agency (AZOP), the State Election Commission (DIP), the Croatian Academic and Research Network (CARNet), the Croatian Bureau of Statistics, the State Audit Office, the Croatian Institute for Social Work, and the National Foundation for Civil Society Development. The Central Finance and Contracting Agency (SAFU) stands out as the only institution that reported neither using nor testing AI, while nevertheless already preparing rules governing its use.
Which Institutions Use the Largest Number of Different AI Solutions?
Five institutions (HZMO, HZZ, HZZO, the Ministry of Health, and HAMAG-BICRO) reported that they apply existing regulations or rules to the use of AI in their work, for example those related to information and cybersecurity. This means that some institutions incorporate the use of AI into existing systems of rules and procedures, rather than establishing separate internal rules for its use.
Finally, 26 institutions reported that they use or test AI solutions in their work without specific internal rules. They include four ministries (the Ministry of the Interior, the Ministry of Economy, the Ministry of Culture and Media, and the Ministry of Regional Development and EU Funds), 10 courts, the Agency for Electronic Media (AEM), the State Geodetic Administration, HERA, three cities (Osijek, Rijeka and Split), two counties (Primorje-Gorski Kotar and Split-Dalmatia), and four clinical hospital centres.
Only One Institution Did Not Respond
Three institutions did not disclose the names of the AI solutions they use, explaining that disclosing this information could compromise their security (HZMO, the City of Rijeka and AZOP). The Ministry of the Interior, meanwhile, did not provide us with the name of the manufacturer of its Face Recognition system for facial recognition and biometric identification, and we were unable to obtain this information from publicly available sources either.
The ministry also stands out because we found that it uses systems whose application may pose particular risks to human rights and fundamental freedoms. These include three AI systems for biometric identification and security. In addition to Face Recognition, they are ABIS (Automated Biometric Identification System) and NABIS (National ABIS System).
The only institution that did not respond to Gong’s standardized questionnaire was the State Inspectorate.
The findings of Gong’s research empirically demonstrate that a public register, clear national standards and consistent internal rules should become an integral part of AI governance in Croatia’s public sector, not as an obstacle to its use, but as a prerequisite for transparent and responsible use while protecting human rights.
An Opportunity to Set Standards
The differences in the responses show that institutions do not have a uniform way of recording AI solutions, that some have an insufficient understanding of what constitutes the use of AI, and that they have differing views on the need for specific rules governing the use of AI in their work. As a national framework for monitoring the use of AI in the public sector is being established, there is an opportunity to set standards before inconsistent practices become entrenched, with some institutions developing their own rules while others are not considering the issue at all.
Gong therefore recommends establishing a public, unified and regularly updated register of AI solutions used by public authorities, together with clear national standards for recording them and easily understandable information for citizens about their use. Public authorities should also adopt clear internal rules for the responsible use of AI and invest in building employees’ knowledge of the capabilities and limitations of these systems, the protection of human rights and data, and accountability for decisions in which AI is involved.
