Serbia - AI Ministerial Declaration

Overview

The Belgrade Ministerial Declaration on Artificial Intelligence was adopted on December 4, 2024, at the 2024 GPAI Summit in Belgrade, Serbia. This international declaration represents the collective commitment of 44 member nations to advancing safe, secure, and trustworthy AI development. The adoption was announced by Serbia's Ministry of Science as a major diplomatic success, positioning Serbia as a leader in global AI governance. The declaration builds on the OECD AI Principles and continues GPAI's mission of promoting responsible AI through international cooperation.

Definitions

The declaration adopts definitions consistent with the OECD AI Principles framework. Artificial Intelligence encompasses systems capable of learning, reasoning, and decision-making to achieve objectives. Safe AI refers to systems designed to prevent unintended harm throughout their lifecycle. Secure AI addresses protection against malicious manipulation and unauthorized access. Trustworthy AI encompasses systems that are safe, secure, respect human rights, and operate transparently with appropriate accountability. Human-centric AI prioritizes human well-being, autonomy, and rights in system design and deployment. The AI system lifecycle covers development, deployment, operation, and retirement phases.

Governance and Institutional Framework

The declaration operates within the Global Partnership on Artificial Intelligence (GPAI) governance structure, now integrated with the OECD. Serbia and Slovakia jointly chair GPAI following the Belgrade Summit, providing strategic leadership for the partnership. The GPAI Ministerial Council serves as the supreme decision-making body, with participating members including governments, industry, academia, and civil society. Working groups address specific AI policy areas including responsible AI, data governance, innovation and commercialization, and future of work. The Serbia Office for Artificial Intelligence coordinated national hosting and contributed substantively to declaration development. UNDP and OECD provided technical support throughout the summit process.

Key Focus Areas

• Safe AI Development: Commitment to ensuring AI systems are designed with safety considerations throughout their lifecycle.
• Secure AI Systems: Promoting cybersecurity measures and protection against AI system manipulation.
• Trustworthy AI: Advancing AI that operates transparently, respects rights, and maintains accountability.
• Human-Centric Approach: Prioritizing human well-being, autonomy, and dignity in AI development.
• Democratic Values: Preserving individual liberty and democratic principles in AI governance.
• Rule of Law: Ensuring AI development and use operate within legal frameworks.
• Human Rights Protection: Safeguarding fundamental rights in AI applications.
• Ethical AI Development: Rooting AI advancement in established ethical principles.
• International Cooperation: Strengthening multilateral collaboration on AI governance.
• Knowledge Sharing: Public dissemination of GPAI's work for global benefit.
• Inclusive Development: Ensuring all nations can benefit from AI advancements.
• Stakeholder Engagement: Involving ministers, organizations, academia, industry, and civil society.

Implementation Framework

Implementation operates through the GPAI multi-stakeholder structure. Member countries commit to aligning national AI policies with declaration principles. GPAI working groups develop practical guidance and tools supporting implementation. Technical experts from academia and industry contribute research and best practices. National governments maintain primary responsibility for domestic implementation while coordinating internationally. The co-chairs (Serbia and Slovakia) facilitate ongoing dialogue and progress monitoring. OECD provides technical support and analytical resources. The declaration connects to the broader OECD AI Principles implementation framework, leveraging existing mechanisms and monitoring processes. Future ministerial meetings will review progress and update commitments as needed.

Monitoring and Evaluation

Progress monitoring occurs through GPAI's established mechanisms. The Ministerial Council convenes regularly to assess implementation and address emerging issues. Working groups report on specific policy area developments. National reports contribute to aggregate progress assessment. The OECD.AI Policy Observatory tracks AI policy developments across member countries. Academic and industry experts provide independent evaluation through GPAI participation. The summit commitment to meet in Bratislava under continued Serbia-Slovakia leadership ensures ongoing accountability. International AI governance events, including the February 2025 Paris AI Action Summit, provide additional review opportunities. Public dissemination of GPAI's work supports transparency and external scrutiny.

Penalties, Liability, and Appeals

As an international ministerial declaration, the Belgrade Declaration operates through diplomatic commitment rather than binding legal obligations with direct penalties. Non-compliance may affect a country's standing and influence within GPAI and broader international AI governance forums. Peer review and public reporting create reputational incentives for adherence. National implementation requires domestic legal frameworks to establish enforceable requirements. Member countries retain sovereignty over specific implementation approaches and penalty structures. The declaration's voluntary nature encourages broad participation while respecting diverse legal traditions. Dispute resolution follows diplomatic channels within the GPAI and OECD frameworks.

Relationship to Other Instruments

The declaration explicitly builds on the OECD AI Principles adopted in 2019, extending their reach through GPAI membership. It connects to previous GPAI declarations including the Tokyo Declaration (2023) and Montreal Declaration (2022). The declaration acknowledges the upcoming AI Action Summit in Paris (February 2025), indicating coordinated international efforts. At national level for Serbia, it complements the AI Development Strategy and Ethical Guidelines. The EU AI Act and national AI legislation in member countries provide binding frameworks aligned with declaration principles.

International Alignment

The Belgrade Declaration represents strong international consensus on AI governance principles. With 44 countries adopting the declaration, it demonstrates broad multilateral support for human-centric AI. Alignment with OECD AI Principles ensures consistency with the leading international AI governance framework. The declaration complements regional initiatives including the EU AI Act and national AI strategies. G7 and G20 AI commitments inform and align with GPAI's work. UNESCO's AI Ethics Recommendation principles resonate throughout the declaration's human rights emphasis. The declaration's acknowledgment of diverse national approaches reflects inclusive international cooperation respecting sovereignty while advancing shared values.

Implementation Timeline

Sources and References