China Releases an Action Plan for Global Artificial Intelligence Governance

On 26 July 2025, during the High-Level Meeting on Global AI Governance held in Shanghai as part of the World Artificial Intelligence Conference, the People’s Republic of China unveiled a strategic document entitled the Action Plan for Global Artificial Intelligence Governance. This release marks a continuation of the diplomatic initiative launched by President Xi Jinping in October 2023 and reaffirms China’s ambition to shape the emerging global legal and normative framework for artificial intelligence.

‍

A Soft Law Instrument with Multilateral Reach

While not legally binding, the Action Plan constitutes a significant soft law instrument. It articulates a set of political commitments and normative orientations aimed at informing future domestic legislation, multilateral agreements, technical standards, and voluntary codes of conduct across the international AI ecosystem.

The Plan is presented as China’s concrete contribution to the United Nations Global Digital Compact, and lays out a vision for international AI governance based on sovereign equality, security, inclusive development, and shared responsibility.

It calls for broad cooperation among key stakeholders, including national governments, international organisations, AI technology companies, academic institutions, and civil society actors.

‍

Foundational Principles of the Proposed Governance Model

The Action Plan is structured around six foundational principles, each reflecting China’s preferred approach to global technological governance:

1. Serving the common good: AI development must benefit human well-being and support social progress, treating AI as a global public good.
2. Respect for state sovereignty: Each country must retain full discretion over its domestic AI policies, including regulation, infrastructure, and data governance.
3. Development for equity and inclusion: AI should not exacerbate inequalities between nations or populations; access to critical AI resources (data, compute, foundation models) should be open and equitable.
4. Safety and controllability: AI systems must be technically auditable, predictable, and under human control, to mitigate risks of misuse and systemic harm.
5. Fairness and universal accessibility: The benefits of AI innovation should be broadly distributed and socially inclusive, particularly for the Global South.
6. Openness and international cooperation: China calls for renewed multilateralism in AI governance, with inclusive participation from developing countries in norm-setting and implementation mechanisms.

Although these principles are framed in general terms, they offer a distinctive normative approach—one that contrasts with both the risk-based regulatory model of the European Union (AI Act) and the market-driven, innovation-first approach of the United States.

‍

Strategic Axes and Thematic Priorities

The Action Plan identifies 13 priority areas for international collaboration. Key among them are:

• The creation of a multilateral coordination mechanism dedicated to AI governance. While no formal institution is named, the document implies support for an international body or platform—possibly headquartered in Shanghai—to facilitate standard-setting, information-sharing, and policy dialogue.
• The development of shared technical and legal standards, particularly in the areas of algorithmic transparency, foundation model safety, risk assessment, and developer/operator accountability.
• The promotion of open and inclusive experimentation platforms, granting developing countries access to AI models, training data, compute infrastructure, and collaborative research environments.
• North-South and South-South cooperation, including technology transfer, skills development, and co-development of AI governance tools tailored to local contexts.
• Voluntary mechanisms for AI oversight and certification, including interoperable conformity assessment schemes and transnational transparency frameworks.

In addition, the Action Plan emphasises the importance of cybersecurity in the AI lifecycle, the prevention of malicious uses (e.g., deepfakes, disinformation), and the establishment of globally shared ethical principles.

‍

Towards a Multipolar Governance Architecture?

The Plan positions itself as a strategic counterproposal to Western-centric governance models. Rather than prescribing legally binding obligations or adopting a human-rights-based framework, China promotes a model of cooperative standardisation based on mutual respect, technological sovereignty, and solidarity with the Global South.

This stands in contrast with:

• The EU Artificial Intelligence Act, which establishes binding legal requirements for high-risk AI systems, including conformity assessments, documentation duties, and market surveillance mechanisms;
• The Council of Europe’s international AI treaty (signed in September 2024), which grounds AI governance in the rule of law, democratic values, and the protection of fundamental rights.

The Chinese approach instead favours dialogue-based governance, voluntary harmonisation, and inclusive participation over ex ante regulatory intervention.

‍

Conclusion

The publication of this Action Plan underscores China’s ambition to play a leading role in shaping global artificial intelligence governance. It presents a normative framework designed to structure future multilateral cooperation and to position AI as a global common good governed through shared principles of equity, sovereignty, and collective benefit.

As a French-qualified lawyer, I emphasise that this document is not legally binding under French or European law. For any detailed analysis regarding its implications under Chinese law or within the Asia-Pacific region, consultation with a qualified Chinese legal expert is strongly recommended.