Law on Artificial Intelligence to be adopted by the end of the year

The deadline for adopting the law on artificial intelligence is December this year, Marija Gnjatovic, State Secretary at the Ministry of Science, Technological Development and Innovation, told Radio Belgrade. She added that Serbia was not late in the process of adopting such a document and that the goal was to protect the safety and health of citizens while supporting the innovation ecosystem and startup companies.

Serbia is preparing the first law on artificial intelligence that will be fully aligned with European standards, and the deadline for its adoption is December this year, said Marija Gnjatovic, State Secretary at the Ministry of Science, Technological Development and Innovation, on Radio Belgrade.

She emphasized that the adoption of the new regulations would not bring any significant changes in terms of the availability of models and systems in everyday life, but that, rather, the main goal of the law was to protect citizens from misuse.

Transparency and labeling of AI content

Speaking about informing citizens, Gnjatovic emphasized that under domestic law, it would be absolutely mandatory for citizens to be informed when viewing a photo, video, or listening to an audio recording created or modified by AI tools.

However, she emphasized that the law and content labeling were not enough, but that the implementation of regulations had to go hand in hand with educating the population in order to raise awareness of how models and systems made decisions.

Different levels of risk and role of humans

The law does not regulate the technology itself, which is impossible to predict in this way, but regulates its application, which carries different levels of risk.

• High risk: Application of artificial intelligence systems in areas such as healthcare, for decision-making or high levels of assistance.
• Low risk: Use of open models and artificial intelligence to recommend content, movies, or music, where it is only important to inform the user that the application of AI has led to certain solutions.

Commenting on the use of artificial intelligence in the media and science, Gnjatovic pointed out that times had changed and that now the world’s leading publishers were engaged in education on how AI was used for the best publications.

– The basic principle of both the European act and our law is that humans remain at the center of decision-making - as the central figure in it – said Gnjatovic.

Implementation control and institutional framework

When it comes to monitoring the implementation of the law, the State Secretary stated that the institutional framework was still developing worldwide, because technology was developing faster than the institutions reacted.

According to predictions, supervision will be carried out by the market inspection, while the responsibility lies with the manufacturers and users of AI systems. For high-risk systems, it will be mandatory to submit documentation on how the accuracy and effects of the software are monitored in the future. It is also planned to form a regulatory body that will maintain a register of systems and models that are applied on the Serbian market.

Gnjatovic emphasized that Serbia was not late in this process, but that through international cooperation it was preparing an exceptionally high-quality document whose primary focus was the protection of the safety and health of citizens, but while simultaneously supporting the innovation ecosystem and startup companies.

Ministarstvo nauke, tehnološkog razvoja i inovacija Republike Srbije

Ministry of Science Technological Development and Innovation

Marija Gnjatović

Law on Artificial Intelligence