International AI Ethical Governance Action Plan

Artificial Intelligence (AI) is a strategic technology leading a new round of scientific and technological revolution and industrial transformation. It opens up new prospects for human society while posing ethical challenges in science and technology that cannot be ignored. To promote the well-being of humanity, foster the peaceful and sustainable development of human society, and advance an inclusive, safe, trustworthy and sustainable AI future for all, we hereby put forward the International AI Ethical Governance Action Plan, calling on all parties to take joint action, build consensus on ethical governance, and fully deliver on the commitments set out in the United Nations Pact for the Future and its annex, the Global Digital Compact, so as to create an intelligent future where the benefits of technological innovation are universally shared, governance mechanisms are transparent, trustworthy, open and comprehensive, and where free and all-round development is achievable and within reach for all.

We reaffirm that the development and use of artificial intelligence must be grounded in international law, including the Charter of the United Nations, and must respect, protect and promote human rights and fundamental freedoms. AI systems should be designed, deployed and used throughout their life cycle in a manner that upholds the right to privacy and data protection, ensures human oversight and accountability, and remains consistent with the 2030 Agenda for Sustainable Development. Particular attention should be paid to the rights and interests of vulnerable groups, including women, children, older persons, persons with disabilities, workers, youth, micro-, small and medium-sized enterprises, and enterprises in developing countries.

I. Promoting Broad Participation of All Parties in Ethical Governance

We uphold the global governance philosophy of extensive consultation, joint contribution and shared benefits, and call for an ethical governance framework led by governments, facilitated by exchanges and mutual trust among international organizations, advanced by innovation-driven enterprises and research institutions, and with the active participation of civil society organizations and individuals. We advocate the establishment of cross-country, cross-regional, cross-sectoral and cross-disciplinary collaboration mechanisms for AI ethical governance. We support the United Nations in playing a key role in global AI governance, enhance the representation and voice of developing countries in global AI governance, and jointly safeguard the equal right to develop and use AI.

II. Strengthening Ethical Governance Throughout the AI System Life Cycle

We call on AI industry actors to pay attention to ethical risks, promote the shift of ethical governance from after-the-fact measures to the very beginning of research and development, and guide industry actors to uphold the principle of fairness and non-discrimination, and to integrate ethical governance into the entire industry life cycle of AI scientific research, technology development, product deployment, operation and maintenance. In the stage of data collection and algorithm design, enhance data authenticity and accuracy, and enhance the transparency and explainability of AI systems. In the stage of AI system deployment and use, we should prevent misuse, abuse and malicious use of the technology.

III. Deepening Cooperation on Ethical Risk Prevention and Control

We encourage countries to explore the establishment of categorized and tiered risk identification mechanisms for ethical risks, taking into account the technical characteristics, application scenarios, scope of impact and potential consequences of AI systems. We need to pay close attention to major ethical risks posed by frontier AI systems, and explore the establishment of a mechanism for graded and categorized risk notification, assessment and dynamic updating. We should jointly explore the conduct of regular categorized and tiered ethical risk assessments, and update relevant reference standards for such risk categories and tiers as appropriate. We call on countries to focus special attention on new forms of AI such as AI agents and embodied AI, and to formulate differentiated risk-handling rules based on factors including data sensitivity, tool invocation permissions, and autonomous decision-making capacity. We also encourage the open source ethics community to strengthen ethical governance and guard against ethical risks associated with open source.

IV. Establishing Dynamic Ethical Governance Mechanisms for Agile Governance

We need to proactively adapt to the rapid iteration of AI systems and the growing autonomous capabilities of AI agents, and implement agile governance. We encourage the establishment of an AI ethical risk case repository and an information exchange mechanism for major ethical risk incidents, accompanied by timely updates to the repository and dynamic optimization of the exchange mechanism, so as to provide an empirical foundation for agile governance. While fully respecting the national sovereignty, security concerns, laws and regulations of all countries, we call for voluntary sharing of risk warnings, response experience and governance tools based on agreed standards and procedures, to jointly enhance the agile governance capabilities of all countries.

V. Cultivating a Collaborative Ethical Governance Ecosystem Across the Industrial Chain

We call for focusing on service provision while strengthening governance, and cultivating a global AI ethical governance ecosystem covering the full chain of research, development, application and promotion. We encourage governments to clarify the roles, responsibilities and boundaries of foundation model providers, AI agent developers, deployers, tool and interface providers, terminal manufacturers, distribution platforms, operation and maintenance entities, and users. We encourage the development of ethical governance technology tools such as explainability, robustness, fairness, reliability, privacy protection, and bias mitigation, and also promote the joint construction and sharing of AI ethical governance datasets and the open-sourcing of governance technology tools.

VI. Fostering a Social Environment for Ethical Governance

We need to launch international cooperation actions to enhance ethical governance capacity, with special emphasis on supporting the ethical governance capacity-building of developing countries. We call for strengthening personnel training for the ethical governance of science and technology, while advancing national education in this field, incorporating the ethics of science and technology into higher education curricula, and cultivating interdisciplinary talents with both technical skills and ethical awareness. We encourage the establishment of platforms for public education and awareness-raising on ethical governance, leveraging the role of social organizations and mass media, to jointly enhance public ethical awareness and the development of a culture of industry self-discipline. We should respect, protect and promote the rights and interests of women, children, older persons, persons with disabilities and other groups, and strive to bridge the digital divide.

The future of artificial intelligence is shaped by all of humanity, and the path of ethical governance in science and technology requires all parties to walk together. We advocate that ethics-first be established as a governance philosophy, and that ethical self-discipline be internalized as a pursuit in action. We strive to build a safer, more trustworthy, fairer and more inclusive global AI ethical governance system, to ensure that AI technologies benefit all humanity and contribute to an inclusive, safe, trustworthy and sustainable AI future for all.