AI regulations in Serbia have evolved significantly since 2019, when the country established its first national AI strategy for 2020-2025. Serbia currently ranks 57th out of 193 countries in the AI Readiness Index, making it a leader in artificial intelligence development within the Western Balkans region.
Additionally, the regulatory landscape is rapidly transforming as Serbia develops comprehensive governance frameworks aligned with EU standards. We’ve observed that the Serbian government has taken decisive action by establishing the Council for Artificial Intelligence to oversee policy development and coordinate implementation across various sectors. Furthermore, preparations are already underway for a new AI development strategy covering 2024-2030, which aims to address critical concerns about data protection while promoting ethical AI applications.
In this guide, we’ll explore how businesses can navigate Serbia’s evolving AI regulations, understand compliance requirements for 2025, and align their operations with both local laws and broader European standards. Whether you’re already operating in Serbia or planning to expand there, understanding these regulations is essential for sustainable business growth in the AI sector.
Serbia’s National AI Strategy and Institutional Foundations
Serbia’s strategic approach to artificial intelligence has been shaped through carefully structured national frameworks that establish the country’s vision for AI development. The government has created a foundation for sustainable growth in this sector through comprehensive strategies and dedicated institutions.
AI Strategy 2020–2025: Economic Growth and Education Goals
The first Serbian AI strategy established clear priorities for integrating artificial intelligence into the national economy. This strategy primarily focuses on boosting economic development through AI implementation across key sectors. Education forms another critical pillar, with emphasis on building human capacity in AI disciplines. The strategy outlines concrete steps for fostering research environments and developing AI-specific educational programs at various levels.
One notable aspect is the strategy’s approach to developing both technical infrastructure and regulatory frameworks simultaneously, creating a balanced ecosystem where innovation can thrive within appropriate legal boundaries. This dual approach distinguishes Serbia’s strategy from those focusing predominantly on either technical advancement or regulatory constraints.
AI Strategy 2025–2030: Reliable and Responsible AI Focus
Looking ahead, the 2025-2030 strategy shifts focus toward ensuring AI systems are both reliable and ethically responsible. This evolution reflects global trends in AI governance, where initial emphasis on development gradually gives way to concerns about safety and ethical implementation.
The second strategy places greater emphasis on creating trustworthy AI systems that align with human values and social expectations. Consequently, this framework introduces more stringent requirements for transparency, accountability, and human oversight in AI systems deployed across Serbia.
Role of the Institute for AI Research and Development
At the heart of Serbia’s institutional infrastructure stands the Institute for AI Research and Development. This organization serves as the central hub for coordinating research initiatives, supporting innovation, and advising on policy matters related to artificial intelligence.
The Institute plays a crucial role in bridging the gap between academic research and practical implementation, essentially acting as the operational arm of Serbia’s strategic vision for AI. Moreover, it facilitates international cooperation, connecting Serbian AI researchers and businesses with global networks and knowledge-sharing platforms.
Legal Framework and the Upcoming AI regulations in Serbia
The legislative landscape for artificial intelligence in Serbia is currently in a transformative phase. Despite being a regional leader in AI development, the country still lacks comprehensive legal regulation specifically addressing AI technologies.
Status of the Draft AI Law and Working Group Activities
Currently, Serbia awaits the adoption of its AI regulations in Serbia, although significant progress has been made in laying the regulatory groundwork. A dedicated Working Group for drafting the AI Law was established in early 2024, comprising representatives from various government bodies, scientific institutions, law firms, and businesses involved in artificial intelligence. The Working Group aims to present the draft law by March 31, 2025. This timeline aligns with Serbia’s broader vision of having the AI Law fully implemented by the end of 2027.
Influence of the EU AI Act on Serbian Legislation
The European Union’s Artificial Intelligence Act serves as the primary model for Serbia’s upcoming legislation. This alignment is particularly relevant considering Serbia’s EU accession aspirations and the extraterritorial application of EU regulations. Indeed, under certain circumstances, the EU AI Act provisions may apply to Serbian entities even before domestic legislation is finalized. This approach reflects Serbia’s practical strategy of maintaining compatibility for businesses operating across both EU and Serbian markets.
Council for Artificial Intelligence: Oversight and Coordination
In parallel with the legislative developments, the Serbian government has established the Council for Artificial Intelligence. This body is tasked with:
- Aligning and coordinating activities related to implementing the strategic AI framework
- Monitoring the implementation of planned measures and activities
- Providing advisory input through proposals, recommendations, and standards
- Organizing and overseeing the preparation of AI-related laws and regulations
The Council works closely with international partners, particularly the Global Partnership on Artificial Intelligence, where Serbia holds the presidency for 2025-2027. This leadership role culminated in the Belgrade Ministerial Declaration on Artificial Intelligence, endorsed by 44 member states and the EU in December 2024.
Ethical Guidelines for Reliable and Responsible AI
In March 2023, the Serbian government formally adopted Ethical Guidelines for the Development, Implementation, and Use of Reliable and Responsible AI. These guidelines, though not legally binding, establish preventive mechanisms to ensure AI aligns with the highest ethical and safety standards.
Explainability, Human Dignity, and Non-maleficence Principles
The ethical framework rests on four fundamental principles. First, the explainability principle requires that all AI processes—from development to decommissioning—remain transparent. Second, the dignity principle emphasizes that AI systems must respect human integrity and never subordinate humans to machine functions. Third, the non-maleficence (do no harm) principle mandates that AI systems incorporate safety mechanisms to prevent damage to persons and property. Finally, the fairness principle protects the rights of all users, with special attention to vulnerable groups.
Conditions for Compliance: Human Oversight and Data Protection
To meet these ethical standards, developers must satisfy seven core conditions. Primarily, human oversight and control must be maintained throughout the AI system lifecycle. Technical reliability and security form the second condition, ensuring systems operate as intended. Third, privacy and personal data protection mechanisms must be robust. Other essential conditions include transparency, diversity, non-discrimination, social and environmental well-being, and clear accountability frameworks.
Technical and Non-Technical Methods for Risk Mitigation
Serbia’s approach to responsible AI implementation combines both technical and non-technical methods. Technical methods include comprehensive testing protocols before deployment, ongoing performance monitoring, regular audits, and established correction procedures. These are presented as recommendations rather than mandates.
Meanwhile, non-technical methods take the form of a self-assessment questionnaire designed to evaluate AI systems against the established principles. Developers are encouraged to complete this questionnaire during the earliest planning stages, helping identify areas for improvement before implementation. Notably, the guidelines specify that high-risk AI systems require separate analysis due to their potential impact on human integrity.
Business Compliance Essentials for 2025
Serbian businesses must prepare for comprehensive AI compliance requirements as the regulatory landscape evolves toward full implementation in 2025. Understanding these requirements will be crucial for organizations developing or deploying AI technologies. Companies implementing AI systems can benefit from specialised legal support through our IT & Technology Law practice.
Risk Classification Based on OECD Framework
Serbia has adopted the OECD’s AI system classification framework to categorize applications according to their risk levels. This approach enables targeted regulation that addresses high-risk applications while fostering innovation in lower-risk scenarios. The classification system includes four risk categories similar to the EU AI Act: unacceptable risk, high risk, limited risk, and minimal risk. Key categories requiring special attention include biometric categorization systems, generative AI, AI embedded in critical infrastructure, and computer-implemented inventions.
High-Risk AI Systems: Evaluation and Monitoring Requirements
High-risk AI applications receive enhanced regulatory scrutiny under Serbia’s framework. These systems must undergo comprehensive evaluation processes assessing fundamental rights impact, risk mitigation measures, and compliance with ethical guidelines. The evaluation includes:
- Thorough risk assessments before deployment
- Implementation of monitoring mechanisms throughout system lifecycle
- Regular audits of system outcomes
- Established correction procedures for identified issues
Data Protection and DPIA Obligations for Businesses
A Data Protection Impact Assessment (DPIA) is mandatory before processing when new technologies like AI might pose high risks to individuals’ rights. The DPIA must include planned processing activities, necessity and proportionality assessments, risk evaluations, and technical and organizational safeguards. This requirement applies especially to AI systems that perform profiling, process special data categories, or conduct systematic monitoring of public areas. Businesses deploying high-risk AI systems must conduct a DPIA in accordance with Serbia’s data protection rules. For more information on data processing obligations and regulatory requirements, please visit our Privacy & Data Protection practice.
Alignment with EU AI Act for Cross-Border Operations
For cross-border operations, Serbian businesses should note that starting August 2, 2025, the EU’s layered supervisory framework takes effect. Providers of General-Purpose AI models exceeding the systemic risk threshold (10²⁵ floating point operations) will face stricter requirements. Companies must implement transparent documentation processes and copyright compliance measures to remain competitive in European markets.
Conclusion
Serbia stands poised at the intersection of technological advancement and regulatory prudence as the country progresses toward comprehensive AI governance. Throughout this article, we have traced the evolution of Serbia’s approach to artificial intelligence regulation, starting with its inaugural strategy and extending to the forthcoming framework expected to fully materialize by 2027.
Serbian businesses must certainly prepare for the regulatory changes coming in 2025. The adoption of OECD-based risk classifications, combined with specific requirements for high-risk systems, creates a structured environment where innovation can flourish while maintaining appropriate safeguards. Meanwhile, the Data Protection Impact Assessment obligations serve as critical tools for responsible AI deployment.
The Serbian government, through its Council for Artificial Intelligence and the Working Group for drafting the AI Law, demonstrates a clear commitment to balanced regulation. Their work aligns domestic frameworks with EU standards, therefore creating a seamless operational environment for companies working across markets.
Perhaps most significantly, the Ethical Guidelines establish a value-based foundation for all AI development within Serbia. These principles – explainability, human dignity, non-maleficence, and fairness – reflect a thoughtful approach that places human welfare at the center of technological advancement.
Companies operating in Serbia should accordingly view compliance not merely as a regulatory burden but as an opportunity to build trust with customers and stakeholders. After all, businesses that embrace these ethical frameworks early will likely gain competitive advantages as the market increasingly values responsible AI practices.
The path forward for AI in Serbia combines ambition with responsibility. As the regulatory landscape continues to evolve, businesses that stay informed and proactively adapt their practices will thrive in this emerging ecosystem. Serbian AI regulation ultimately seeks balance – encouraging innovation while ensuring technology serves humanity’s best interests.
The information contained in this document does not constitute legal advice on any specific matter and is provided solely for general informational purposes.
