The world has been rapidly changing during recent years, and the artificial intelligence (AI) revolution has been contributing to the establishment of new foundations upon which humanity will develop and exist in the future.
The use of AI systems facilitates and accelerates task performance, increases efficiency, and transforms economic functioning, enabling scientific advancement and social change. However, the development, implementation, and use of AI also create concerns and entail certain risks, ranging from potential misuse and security issues to ethical and legal challenges. In this context, the way AI is used and the purposes it serves will shape the future.
Lawmakers around the world are facing complex challenges related to regulating artificial intelligence. Nevertheless, some countries and international organizations have already addressed this issue to a certain extent, adopting laws, recommendations, guidelines, and other legal acts. A legal framework is essential to define measures and activities for the development, implementation, and monitoring of AI systems.
In 2019, the Republic of Serbia laid the foundations for the field of artificial intelligence by adopting the Strategy for the Development of Artificial Intelligence in the Republic of Serbia for the Period 2020–2025, establishing goals and measures for AI development [1]. However, when it comes to AI, this represented only the first step toward the activities to be undertaken in the years ahead.
Below, we will outline and discuss certain acts and regulations relevant to the field of artificial intelligence in the Republic of Serbia. Nevertheless, some of these are still pending.
Artificial Intelligence Act of the Republic of Serbia
Comprehensive legal regulation of artificial intelligence is still awaited, as the Artificial Intelligence Act of the Republic of Serbia has not yet been adopted. However, the importance and necessity of enacting this Law have been recognized, and significant steps have been taken through the establishment of a Working Group responsible for drafting the Artificial Intelligence Act of the Republic of Serbia[2].
While we are waiting for the national AI Act to be adopted, attention should be given to the content of the European Union Artificial Intelligence Act (EU AI Act),which has entered into force and is being implemented gradually. It is believed that the EU AI Act will serve as a model for drafting the Serbian AI Act. Under certain circumstances, the EU AI Act is also relevant for entities outside the European Union, as its provisions may apply extraterritorially. More on this can be found in the text „How to comply with the EU AI Regulation?”.
Development strategies for Artificial Intelligence in the Republic of Serbia
As mentioned, in 2019, Serbia adopted the Strategy for the Development of Artificial Intelligence for the Period 2020–2025[3], and as a continuation of this initiative, in January 2025, a new Strategy for the Development of Artificial Intelligence for the Period 2025–2030[4] was adopted.
1. Strategy for the Development of AI in the Republic of Serbia for the period 2020–2025
This Strategy established the goals and actions for the development of artificial intelligence. Its general goal was “the use of artificial intelligence in order to support economic growth, employment, and a better quality of life.” In addition to this overarching goal, specific objectives were defined, including:
- Development of education aligned with the needs of modern society and economy, determined by AI progress;
- Development of science and innovation in AI and its implementation;
- Development of an AI-based economy (where AI is the key component or where it is implemented across various industries);
- Improvement of conditions for AI development and public sector services through AI;
- Ethical and safe implementation of AI.
A broad set of measures was implemented to support the development of artificial intelligence. The Institute for Artificial Intelligence of Serbia was established as a result of the implementation process of all the strategic measures[5]. Moreover, since ethical and safe use of AI was identified as one of the five specific goals, in 2023, the Ethical Guidelines for the Development, Implementation, and Use of Reliable and Responsible Artificial Intelligence were adopted, which will be discussed later in this text.
2. Strategy for the Development of Artificial Intelligence for the period 2025–2030
At the beginning of 2025, the Republic of Serbia adopted a new Strategy for the Development of Artificial Intelligence for the Period 2025–2030, which defines new goals and measures for the continued development and implementation of AI in the country.
Unlike the previous Strategy, the new one emphasizes the importance of ensuring that artificial intelligence is reliable and responsible. Accordingly, its general goal is “the use of reliable and responsible artificial intelligence in order to support economic growth, employment, and a better quality of life.”
The new Strategy also specifies special objectives and corresponding measures for achieving the general goal:
Specific Objectives | Measures |
1. Establishing and harmonizing the institutional and legal framework for the development and safe, secure, reliable, and responsible implementation of artificial intelligence. | · Improving the legal framework to enable the development and implementation of artificial intelligence. · Enhancing the institutional framework for artificial intelligence. · Continuously harmonizing ethical principles and the legal framework with international standards and ensuring their implementation. |
2. Enhancing and facilitating technological development in the field of artificial intelligence and AI-based solutions. | · Financing and developing scientific research projects in the field of artificial intelligence. · Supporting innovative companies in developing artificial intelligence. · Supporting the further development of the Institute for Artificial Intelligence of Serbia. · Providing incentives for the implementation of artificial intelligence in scientific research. |
3. Developing human resources and knowledge in the field of artificial intelligence. | · Developing qualification standards, curricula, and study programs in the field of artificial intelligence in higher education. · Developing AI-related training in primary and secondary education and through other service providers. · Increasing general AI literacy in society. · Providing non-formal education for the use of AI-based solutions. · Providing non-formal education and recognition of prior learning for IT professionals involved in the design and development of AI-based solutions. · Supporting talent development and providing career guidance and advisory services in the field of artificial intelligence. |
4. Improving infrastructure and resources necessary for the development of artificial intelligence. | · Expanding hardware infrastructure. · Building and connecting ecosystems with participants in the Republic of Serbia and abroad. · Implementing a national program for the development of language technologies for the Serbian language. |
5. Using data as a key resource for the development of artificial intelligence. | · Building capacity for data management and data engineering in public administration. · Opening and using large datasets for artificial intelligence purposes. |
6. Increasing the implementation of artificial intelligence in priority sectors of society and the economy. | · Providing incentives for the implementation of artificial intelligence in the public sector. · Providing incentives for the implementation of artificial intelligence in public administration. · Providing incentives for the use of artificial intelligence in educational institutions. · Providing incentives for the implementation of artificial intelligence in the economy. · Encouraging the development of information security solutions that utilize artificial intelligence. · Promoting the use of artificial intelligence in defense technologies for the needs of the defense system of the Republic of Serbia. · Providing incentives for the implementation of artificial intelligence in healthcare and biotechnology. · Providing incentives for the implementation of artificial intelligence in agriculture. |
Setting these goals and implementing the corresponding measures is crucial for the direction in which the AI will develop. The Strategy encourages innovation, scientific research, technological development, and education in the field of AI, while emphasizing the importance of ethical standards and principles.
Ethical Guidelines for the Development, Implementation, and Use of Reliable and Responsible Artificial Intelligence
As stated earlier, the 2020–2025 Strategy aimed to introduce preventive mechanisms to ensure that AI is implemented ethically and safely. Recognizing the numerous challenges posed by AI development and implementation, it was highly necessary to ensure a responsible way to align AI with ethical and safety standards.
Accordingly, one of the measures was to ensure responsible AI development in accordance with international ethical standards[6]. It was recognized that a balance must be established between promoting AI development on one hand and protecting fundamental human rights and ethical principles on the other. The focus was placed on the proper use and development of safe, reliable, and responsible AI.
As a result, in 2023, the Ethical Guidelines for the Development, Implementation, and Use of Reliable and Responsible Artificial Intelligence (“Ethical Guidelines”) were adopted, incorporating principles from UNESCO’s 2021 Recommendation on the Ethics of Artificial Intelligence.
Although the Guidelines are not legally binding, they serve as a roadmap for all stakeholders in the AI ecosystem, guiding them to minimize, mitigate, or eliminate potential risks associated with AI development and its implementation. The preventive nature of the Guidelines contributes to human rights protection, damage prevention, and strengthening public trust in AI safety.
The Guidelines emphasize the centrality of humans in relation to AI development. Their goal is not to hinder scientific progress but to ensure that technological development proceeds without endangering human dignity, animals, or the environment.
Before outlining the specific conditions, we will first explain in more detailed way what “reliable and responsible” AI means, as defined in the very title of the Ethical Guidelines.
a) Reliable and Responsible Artificial Intelligence
According to the Ethical Guidelines, reliable and responsible AI is based on three independent yet complementary components:
- Technical reliability and safety
- Legality – compliance with legal act and regulation
- Ethical alignment – compliance with established ethical principles and values
Fulfilling one of the components does not imply the fulfilment of the others, as each is assessed independently.
Reliability and responsibility in AI are ensured through the balanced interaction of all components, none of which may be omitted. Only when technical reliability, legality, and ethical compliance are all met can AI be considered truly “reliable and responsible.”
Meeting these standards is a continuous process that must accompany the entire life cycle of AI development.
After clarifying the meaning of “reliable and responsible” artificial intelligence, the following section outlines the principles defined in the Ethical Guidelines as the foundation for its creation, development, implementation, and use.
b) Principles
The Ethical Guidelines identify several fundamental principles forming the basis for creating, developing, implementing, and using AI systems:
- Explainability and verifiability – AI must be transparent and understandable at all stages, from development to decommissioning, allowing continuous verification of its operation and impact. Viewed separately, explainability means that all processes must remain transparent — from development, testing, and deployment to system monitoring and decommissioning — while verifiability means that the system can be examined throughout its entire life cycle, including the assessment of its impact on humans in both the near and distant future.
- Human dignity – AI must respect human integrity, autonomy, and freedom, and must never subordinate humans to machines. This principle implies that artificial intelligence should not be developed or used in a way that endangers human beings. Based on the understanding that artificial intelligence is a tool intended to enhance human life, humans should never be subordinated to artificial intelligence.
- Non-maleficence (Do no harm) – AI systems must be safe and prevent harm to people or property, with mechanisms for timely remediation and compensation where necessary. AI systems must comply with safety standards and include mechanisms to prevent harm to people and their property. If any damage occurs, it must be remedied as soon as possible, and the affected individuals must be compensated in accordance with the law.
- Fairness – AI must prevent discrimination, ensure equal opportunities, and provide avenues for contesting and legally challenging AI-driven decisions. This principle implies that the protection of rights and integrity from discrimination, meaning that artificial intelligence systems must prevent discriminatory practices in their use. AI systems should ensure equal opportunities for all, as well as provide the possibility to challenge and seek legal protection against decisions resulting from the operation of AI systems and the actions of those responsible for their functioning.
c) Conditions for Reliable and Responsible AI
To ensure reliability and responsibility, AI systems must meet verifiable conditions corresponding to these principles, including:
- Human oversight and control;
- Technical reliability and safety;
- Privacy and data protection;
- Transparency;
- Diversity, non-discrimination, and equality;
- Social and environmental well-being;
This implies that the principles represent the foundation and the value framework to be pursued, while the conditions consist of verifiable parameters that confirm and demonstrate that the principles have been fulfilled in practice. This is achieved through methods that can be divided into:
- Technical methods – recommendations guiding AI systems toward reliable behaviour;
- Non-technical methods – questionnaires for self-assessment and verification of compliance with ethical and reliability conditions. The questionnaires in the Ethical Guidelines are structured in the form of questions and corresponding answers.
Therefore, the recommendations and questionnaires serve as tools for guiding the development and use of artificial intelligence toward reliability and responsibility. The Ethical Guidelines also allow for additional conditions, provided they align with ethical principles.
Current absence of the AI Act in the Republic of Serbia does not mean that AI operates in a legal vacuum. Contrary, numerous existing legal provisions are already actively applied to the development and use of AI systems – starting from the rules on contractual liability and personal data protection (in particular, the obligation to conduct a Data Protection Impact Assessment – DPIA), through consumer protection law, and up to constitutional and statutory provisions that prohibit discrimination. All these norms together form the basic legal framework governing the field of AI in Serbia, even though a dedicated AI Act has not yet become a part of the Serbian legal framework.
Serbia’s journey toward comprehensive AI regulation is more than a legislative process – it is a step toward defining how technology and society will coexist in the years to come. The anticipated AI Act is expected to be a milestone in achieving that vision, turning strategy and ethics into a binding legal reality.
