The DeepSeek logo is seen in this illustration taken Jan 27, 2025. [Photo/Agencies]

On July 26, the Global AI Governance Action Plan, jointly initiated by China and global partners, was unveiled at the 2025 World Artificial Intelligence Conference and High-Level Meeting on global AI governance.

The plan offers a forward-looking framework for global AI governance, with 13 concrete measures and a systemic design consisting of a governance philosophy, institutional mechanisms and practical pathways.

Clear orientation

The plan's most profound contribution lies in its governance philosophy and repositioning of AI technology as a strategic resource for global development.

Among the six principles of the plan, "aligning AI with development goals" sits at the center. This particular principle signifies a fundamental shift in the paradigm of global AI governance from technology-centric to development-centric, placing priority on narrowing the digital divide and promoting inclusive development.

The plan also puts international cooperation on AI capacity building at the top of the agenda, which not only demonstrates its adherence to the right to development but also represents a profound reflection on the Western-dominated AI governance paradigm.

The principles of "upholding fairness and inclusiveness" and "aligning AI with development goals" reflect the governance philosophy that the plan champions.

A commitment to assisting the Global South in truly accessing and utilizing AI, and promoting AI development in an inclusive and universally-beneficial manner, is also laid out in the plan.

This governance philosophy lays a solid foundation for every subsequent institutional design and practical pathway.

Comprehensive and systemic approach

The plan presents a multi-layered framework that covers technological innovation, industrial application, security governance and international cooperation, with each measure translating the development-oriented governance goals into action.

At the technical level, "creating a diverse, open and innovative ecosystem" and "promoting common understanding on standards and norms" operate as a dual engine. "Cross-border open-source communities and secure, reliable platforms", when fostered, will lower technical barriers and give developing countries a viable path to global AI innovation.

In the meantime, the plan stresses coordination among international standards bodies to prevent any race on inventing new systems to replace the old ones.

On security, the plan expresses the need to "explore categorized and tiered management approaches, and build a risk testing and evaluation system for AI". Instead of the traditional one-size-fits-all regulations, it reflects a nuanced grasp of AI's technical complexity and diversity in application. The proposed traceability-management system for AI services will enable AI lifecycle supervision, and thus ensure a balance between security and development.

Chinese characteristics

The plan puts "strengthening international cooperation on AI capacity building" high on the global AI governance agenda, fostering a new model for international cooperation aimed at capacity building.

Key mechanisms in this model include joint laboratories, mutual-recognition platforms for safety assessment, and education and training programs. They form a complete chain of research cooperation, standards recognition and talent cultivation.

In the realm of data governance, the plan calls for the joint creation of high-quality datasets, and for the construction of a global data-sharing mechanism or platform. The former tackles data quality issues, while the latter addresses challenges arising from data flow. Together, they provide the data required for inclusive AI development.

The plan also proposes that public sectors "spearhead deployment and application", becoming leaders and pacesetters in the application and governance of AI. It paves the way for large-scale AI application and, more importantly, creates favorable conditions for the formation and promotion of governance standards through the demonstration effect of the public sector.

Environmental sustainability is woven throughout the text. By establishing energy and water efficiency standards for AI, along with the exploration and innovation in the modes of resource-saving and environmentally friendly AI development, the plan makes sustainable AI development a viable reality rather than an aspiration.

Ultimately, AI is not measured by the sophistication of its chips, but by its capacity to amplify human potential and bridge divides. With the help of AI, it is hoped that there will be a brighter future to share.

That is what the plan is about.