BRICS countries advocate for a UN-led global AI governance framework, promoting inclusive development, data sovereignty, and technological equity, to support the digital economy growth and risk prevention in Global South nations.
Source: Ministry of Foreign Affairs
Artificial intelligence represents a significant opportunity to drive the world towards a more equitable future, fostering innovation, enhancing productivity, promoting sustainable development practices, and tangibly improving the well-being of all humanity.
To realize this vision, global governance of artificial intelligence must mitigate potential risks and address the needs of all countries, particularly those in the Global South. Governance must operate under the United Nations Charter and national regulatory frameworks, respecting national sovereignty and adhering to principles of representativeness, development orientation, accessibility, inclusivity, dynamic updating, and agility. Governance should prioritize personal data protection, safeguard human rights, ensure safety, transparency, and sustainability, and bridge the growing digital and data divides within and between nations. Countries worldwide should collaborate to build a UN-centered AI governance system, uphold shared values, address risks, build trust, and ensure inclusive international cooperation and accessibility in AI, including strengthening AI capacity-building in developing countries.
At the international level, the proliferation of governance initiatives and divergences in multilateral coordination may exacerbate existing asymmetries and legitimacy deficits in global digital governance, further undermining multilateralism.
To support constructive dialogue and achieve a more balanced governance approach, we, the leaders of BRICS countries, unanimously agree to develop a set of guidelines to responsibly advance the development, deployment, and application of AI technologies, contributing to sustainable development and inclusive growth. These guidelines are limited to non-military AI and should be implemented through domestic or applicable international frameworks, interoperable standards, and agreements in an inclusive, transparent, and consensus-based manner.
1. Multilateralism, Legitimacy, and Digital Sovereignty
The United Nations is the cornerstone of global AI governance.Global AI governance must avoid fragmentation and duplication at all costs. We must rely on the UN system as a fully inclusive and representative international framework to advance AI governance. In decision-making processes related to UN initiatives, we must promote meaningful participation and influence of emerging markets, developing countries, and the Global South, while recognizing the complementary roles of regional and multi-stakeholder frameworks. We encourage AI policy exchanges and dialogues to spur innovation and economic growth.
AI collaborative governance is complex but promising.Stakeholder networks in developed and developing countries should contribute expertise, ideas, and resources according to their roles and responsibilities. Reaffirming the foundational and leading role of governments in AI governance, we are committed to close collaboration with the private sector, civil society, international organizations, research and academic institutions, and other stakeholders to achieve representativeness and inclusivity.
Digital sovereignty and the right to development are key to global AI governance.We firmly support the right of all countries to harness the developmental benefits of the digital economy and emerging technologies, especially AI, while safeguarding fundamental rights. This includes strengthening digital infrastructure, nurturing local technical talent, ensuring citizen protection from AI risks, and developing AI regulatory frameworks within national legal jurisdictions to enhance AI research capabilities, foster indigenous technological innovation, ensure data security, and advance national digital economies.
2. Market Norms, Data Governance, and Technological Accessibility
Fair competition and market norms are the foundation for AI's equitable future.The digital economy should clarify the rights and obligations of states, businesses, and users under domestic legal frameworks, regulations, and applicable international agreements to build a fair competitive market environment that fosters innovation and economic growth. We emphasize the importance of avoiding fragmented regulatory systems and promoting fair, transparent market norms to encourage competition, enhance competitiveness, prevent distortions, and create a sustainable, healthy market environment.
Data governance is key to inclusive AI governance.We stress that fair, inclusive, and equitable data governance is critical for developing countries to leverage emerging technologies like the digital economy and AI. We also recognize the need to establish data governance frameworks under the BRICS Data Economy Governance Memorandum to build trust, providing developing countries with fair, consistent, secure, and reliable high-quality data while respecting privacy rights, personal data protection, algorithmic transparency, intellectual property, and national security.
Access to AI technology should follow principles of fairness, justice, empowerment, and inclusivity.All countries, regardless of their economic development stage, have the right to develop, use, and benefit from AI. We emphasize the need for international cooperation to facilitate access to AI technologies and critical components, remove financial barriers to AI research and innovation, and build necessary knowledge, skills, and risk management frameworks in low- and middle-income countries to effectively utilize AI.
A balance is needed between protecting intellectual property and safeguarding public interests.A balance should be struck between patents, transparency, and accountability to uphold public interests, promote international technology transfer, and comply with domestic and applicable international laws. Appropriate intellectual property protections, especially copyright mechanisms, should be established against unauthorized AI use to prevent exploitative data extraction and privacy violations, alongside fair compensation mechanisms. Such protections should clarify accountability and adhere to legal requirements for transparency in AI model inputs and outputs.
AI development should uphold openness and encourage an innovative ecosystem.In line with national policies and priorities, we encourage open science and innovation mechanisms to drive open-source development and international scientific collaboration as key engines for enhancing AI research, development, innovation, data protection, data sovereignty, and deployment capabilities. This includes enabling researchers, developers, and organizations to review and audit AI systems to build safe, trustworthy, and transparent AI. We must ensure substantive participation and foster inclusive collaboration to prevent barriers in global AI development and supply chains. We encourage the development of resource-efficient, specialized, open foundational models to advance the AI innovation ecosystem.
International standards should promote inclusive, representative, and accessible AI.Technical standards, norms, and protocols for AI systems should involve public sectors, standards organizations, and UN specialized agencies to ensure trustworthiness, interoperability, security, and reliability across the AI development lifecycle and cross-platform applications. We must prevent standards from becoming disguised market access barriers for SMEs and developing economies.
3. Equity and Sustainable Development
AI should benefit all equitably.Sound infrastructure, meaningful connectivity, and digital inclusion are prerequisites for any country to deploy artificial intelligence. Digital government services built on foundations such as digital public infrastructure may serve as catalysts for the inclusive development of the digital economy and the protection of citizens' rights and welfare. Domestic policies and multilateral development banks should address widespread gaps in capacity and infrastructure to advance the application and development of AI in emerging markets and developing countries.
AI drives collaborative sustainable development.We are committed to supporting AI applications through open-source and other means to address key development challenges in healthcare, education, security, transportation, energy, agriculture, environment, water resources, and waste management, with efforts aligned with national laws and priorities. The benefits of AI for sustainable development should be continuously promoted. We will focus on fostering and empowering local technological capabilities and ambitions through research, development, and innovation initiatives to bridge the technological divide between developed and developing countries.
Environmental sustainability is a prerequisite for AI development.AI should help reduce emissions, adapt to climate change, protect the environment, promote sustainable resource management and optimization, maintain ecological balance, and contribute to sustainable economic and social development, in line with national priorities. The development and deployment of AI should be sustainable, minimizing its environmental impact and addressing issues such as greenhouse gas emissions, energy and water consumption, material use, and electronic waste. Developing countries should seize the window of opportunity brought by the transition to a low-carbon economy and lead development through technological advancements in AI.
AI should ensure decent work and enhance productivity.AI systems have the potential to increase productivity, stimulate innovation, and create jobs, but they also pose challenges, concerns, and risks to the labor environment, work intensity, and job displacement, threatening employment and workers' dignity. With the rapid development of digital platforms and the growing structural impact of generative AI on the job market, we must safeguard the rights and well-being of all workers, especially those directly affected by digital transformation. The adaptability and compatibility of human resources should be considered in the design, deployment, and use of AI, leveraging its advantages to promote decent work and achieve full and productive employment.
AI should be a tool to transform education and learning.AI systems can organize information, customize personalized experiences, and provide recommendations. We are concerned that over-reliance on AI systems may affect human cognition, decision-making processes, and the ability to navigate complex information environments. We are committed to enhancing digital literacy, particularly among teachers and students, enabling them to critically evaluate AI-generated content, understand inherent algorithmic biases, and cultivate strong knowledge autonomy and critical thinking skills.
IV. Ethical, Trustworthy, and Responsible AI for the Benefit of All
AI must be inclusive.We recognize the risk of misuse and distortion due to the lack of knowledge, cultural heritage, and cultural values in datasets and AI models. We reaffirm the importance of building ethical, transparent, and accountable frameworks, such as UNESCO's Recommendation on the Ethics of AI. The development of AI systems must fully respect linguistic, cultural, racial, geographic, and demographic diversity. International cooperation should be pursued to train comprehensive, multilingual, and inclusive high-quality datasets and cultivate local AI talent.
Efforts must focus on mitigating discriminatory biases.We need robust tools to identify and reduce errors and negative algorithmic biases, ensure independent audit mechanisms to uphold fairness, and establish assessment criteria for bias risks to avoid discriminatory and exclusionary practices. AI systems trained on discriminatory data often disproportionately affect women, minorities, persons with disabilities, and vulnerable groups such as children, adolescents, and the elderly. Interdisciplinary collaboration among diverse backgrounds is crucial for setting standards, improving model interpretability, and developing practical governance strategies to prevent negative biases and support the responsible and fair development of AI systems.
Public interest must be prioritized.We will uphold a human-centered approach, fostering harmonious human-machine interactions and ensuring AI remains a powerful tool to enhance human capabilities, with ultimate human control and oversight. We will prioritize human oversight mechanisms, enhance AI decision-making transparency, and establish robust accountability systems to advance technology development and application responsibly and safely, minimizing risks and maximizing social benefits.
A fact-based approach must be upheld.AI-generated false text, images, audio, and video content poses serious threats to information authenticity and integrity, potentially enabling舆论manipulation, social unrest, and undermining public trust in institutions. We will adopt a multidimensional governance approach to enhance information integrity, strengthen media education strategies and local communication efforts, including developing tools to quickly identify misinformation and disinformation, improving individual digital literacy and critical thinking to discern online content, and establishing clear ethical guidelines and regulations to balance AI development and application in information dissemination with privacy and data security protections.
AI systems must uphold safety and trust.We recognize the need to achieve safe, ethical, trustworthy, and responsible AI development for the benefit of all humanity. Countries must collaborate to advance AI innovation and accessibility while addressing immediate and long-term risks posed by AI technologies, in line with national policies and security considerations. We reaffirm the importance of jointly addressing and preventing risks related to the malicious use of AI. AI systems and other information and communication technologies should be designed prudently to detect and prevent misuse, such as fraud, cyberattacks, cybercrime, or data manipulation.
V. The Path Forward
General AI should be developed with prudence.It is crucial that general AI research adheres to ethical guidelines and is applied responsibly and trustworthily to support economic growth, particularly in emerging markets and developing economies, and to address pressing socio-economic challenges. If only a few entities control general AI, it could exacerbate inequality, create new technological dependencies, and pose severe challenges to sustainable development.
Toward fair and inclusive AI.BRICS countries will take joint and proactive steps to create a fair digital environment for all, identify common perspectives, strengthen global cooperation, advance the UN’s global AI governance dialogue and the International AI Science Panel, and ensure meaningful participation and influence of developing countries in relevant processes. We will pool strengths to promote common principles, including the above guidelines, on international platforms. We welcome contributions, especially from developing countries, to refine the guidelines and remain open to continuously optimizing these principles.