Imagine a small business owner in São Paulo who exports coffee to China. She uses an artificial intelligence tool to translate product descriptions and respond to foreign customers. One day, a Chinese buyer complains that an AI-generated description has overstated the coffee's organic certification. Who is responsible: the Brazilian exporter, the Chinese sales platform, or the European company that developed the tool?
Ongoing in China from Monday to Thursday, the 2026 World Internet Conference Wuzhen Summit — focused on open-source development and shared benefits — brings a central challenge in AI governance into focus: how can the benefits of AI be shared more widely, and how can developing countries have a meaningful say in shaping its future?
While individual nations have advanced their own frameworks, such as China's Global AI Governance Initiative, the debate has officially shifted on a global scale.
The international community is no longer debating whether AI requires uniform international governance, but rather how to implement that governance in practice.
I am not convinced that the answer is a world authority controlling technological development. Such an institution would be neither desirable nor politically feasible.
Nor should harmonization mean imposing a single regulatory model on societies with different institutions and priorities.
What we need are shared principles, greater compatibility between regulatory systems and reliable channels of cooperation when the effects of technology cross borders.
For businesses, governance serves two functions. It can reduce costs by harmonizing standards and avoiding duplicative compliance requirements.
More importantly, it reduces uncertainty: the risk that rules will suddenly change or conflict across jurisdictions. Companies can calculate known costs and prepare for identifiable risks. What they cannot easily manage is an environment in which the rules themselves are unpredictable.
For a small exporter, hiring teams of lawyers to interpret several AI regimes is not realistic. Avoiding a market may appear safer than entering it. Costs can be calculated; persistent uncertainty prevents investment.
Yet reducing uncertainty is not enough. The deeper problem is the unequal distribution of technological power. Most developing countries lack advanced semiconductor industries, extensive computing infrastructure or globally dominant platforms.
Without deliberate cooperation, they risk entering the AI economy primarily as data providers, consumers of foreign systems and markets for services designed elsewhere. This is already becoming a defining feature of the digital economy.
A system cannot be genuinely global if a few countries and corporations write the rules while everyone else merely provides data, markets and users. Governing AI must therefore go beyond addressing disinformation, discrimination and privacy violations. It must also improve access to computing capacity, knowledge, infrastructure and skills.
Otherwise, regulation may make AI safer while leaving technological dependency untouched.
The Global Governance Initiative offers a useful framework. Its principles of sovereign equality, the international rule of law, multilateralism, a people-centered approach and real actions apply directly to AI.
Sovereign equality should entail meaningful participation in setting international standards, not merely occupying a seat at diplomatic meetings. Multilateralism means that no single power, regional bloc or private platform should determine the technological future alone.
But principles do not implement themselves. This is where the GGI will be tested. For developing countries, practical action must include investment in digital infrastructure, education, institutional expertise and AI models that can operate in local languages and cultural contexts.
Representation without capability is hollow. Countries cannot exercise meaningful digital sovereignty without the knowledge and resources required to make genuine choices.
Digital sovereignty should not be confused with technological isolation. No country can or should produce every component of the digital economy on its own. Sovereignty in the AI era means having the capacity to choose partners, assess risks, protect citizens, and negotiate from a position of knowledge. International cooperation should expand those choices, not force countries into competing technological camps.
One year after the GGI was proposed, its value should be measured not by diplomatic support alone, but by its ability to foster participation, predictability and practical results. The UN's Global Dialogue and the Independent International Scientific Panel on AI are promising beginnings. However, their legitimacy will depend on whether developing countries genuinely influence their agendas and benefit from their work.
The GGI is a necessary step, but it cannot address the deeper structural issue alone. For a while, the 19th century was defined by territorial empires; the twenty-first century is characterized by informational empires. Although the tools of control have shifted — from bayonets to bandwidth — the fundamental principle of asymmetry endures.
What was once seen as a universal leveling field in the digital world has instead become a form of technological colonialism, in which data functions as currency and algorithms serve roles once played by armies.
Digital colonialism operates through what may be called the trinity of control — extraction of raw data, dependency on foreign platforms and infrastructure, and cognitive capture through algorithms that shape what people see, think and buy.
The central question is not simply how AI should be governed, but also whether humanity can create common rules without concentrating power, reduce uncertainty without stifling innovation, and encourage progress without deepening inequality.
The future of AI must be built collectively, not delivered to most of the world as a decision already made by a few.
The author is a senior fellow of the Policy Center for the New South in Morocco and a visiting professor at China Foreign Affairs University.
The views don't necessarily reflect those of China Daily.
If you have a specific expertise, or would like to share your thought about our stories, then send us your writings at opinion@chinadaily.com.cn, and comment@chinadaily.com.cn.
