‘The non-transparent use of AI in public administration is seriously damaging’

This lack of transparency represents a serious risk for citizens’ rights and makes it difficult to build trust in the digital technologies that the state uses. The recent case of merging land registries, in which the use of an automated system resulted in the deletion of data on real owners, shows the concrete consequences of the careless and unprepared introduction of artificial intelligence into public administration.

The media reports that the correction of errors is carried out “manually”, while certain groups of citizens, especially farmers, are exposed to the risk of direct financial damage.

“This case is a simple example that shows that the use of AI systems without clear rules, supervision and responsibility can produce serious legal and financial consequences for citizens. Such incidents must not become an acceptable collateral victim of the digitization of public administration,” warns Duje Kozomara from the Politiscope association. Of particular concern is the fact that at the moment there is no publicly available list of AI systems used by the state and public administration. The absence of transparency prevents effective public control and prevents the media, experts and the interested public from assessing the legality, safety and justification of the application of AI in the public sector.

Politiscope calls on the Government of the Republic of Croatia and the Ministry of Digital Transformation to establish a public registry of all AI systems used by public and state administration, following the example of Slovenia. Unlike the minimum requirements from the Artificial Intelligence Act, the Slovenian register will cover all AI systems in the public sector, not just high-risk ones.

“Establishing mandatory standards of transparency, starting with the public registry of AI systems, is the first and necessary step towards the responsible management of digital technologies in the public sector. State and public administration must be examples of responsible and transparent use of artificial intelligence” – said Duje Prkut from Politiscope.

Provisions on the establishment of a public registry of artificial intelligence should be included in the draft of the Law on the Implementation of the Act on Artificial Intelligence, which, together with the AI ​​Strategy, will be passed. eight months of production in a non-exclusive process. While industry representatives are overrepresented, associations dealing with the protection of human and digital rights are completely excluded from the composition of the working groups. The involvement of civil society is key to ensure that the development and application of artificial intelligence in public administration does not harm the fundamental rights of citizens and the public interest. Politiscope calls on the Government of the Republic of Croatia to ensure complete transparency in the use of artificial intelligence and automated decision-making systems in the state and public administration, and to establish without delay truly participatory processes for shaping the legal and strategic framework for the development of artificial intelligence.