Moving from AI race to AI governance

Editor's Note: During the 2026 World Artificial Intelligence Conference in Shanghai on July 17, the China University of Political Science and Law and UNESCO jointly hosted the Global AI Policy Dialogue and Collaboration Development Forum. The event was attended by representatives of international organizations, government officials, academics and business leaders from around the world. Below are excerpts of speeches delivered by four of the participants, as reported by Li Wei of China Daily.

Competition alone won't secure the future of AI

During a visit to China last year, I was struck by a manuscript from Alex Karp, CEO of Palantir. He argued that the United States has squandered a generation of engineering talent on trivial pursuits like optimizing online advertising and maximizing engagement instead of tackling the world's hardest problems.

But Karp's solution to that misplaced talent is military competition. His "enemy" is not hunger, disease, climate change, or the existential risks posed by artificial intelligence. It is geopolitical rivalry, especially with China. It reflects the mindset of far too many in the US, but I do not share it.

My first trip to China was in 1989. Back then, parts of the country reminded me of the US in the 1950s. But my subsequent visits to the country changed that impression.

Traveling across China last year, I felt that I was witnessing what the US might look like in 2150. China has achieved remarkable progress, just as the US has its strengths. Both countries have much to learn from each other.

That spirit of mutual learning matters because AI presents humanity with both extraordinary promise and unprecedented danger.

It can accelerate scientific discovery, improve healthcare, strengthen economic development, and tackle global challenges. At the same time, powerful technologies can also be used for destructive purposes.

The uncomfortable truth is that we do not know how to control advanced AI.

We cannot ensure that future systems will remain aligned with human interests, yet we are racing to build them anyway. We are creating what may become the most consequential technology since the atomic bomb without fully understanding how to govern it.

That is why forums like the World AI Conference matter. At a time when competition often overwhelms reason, we need more dialogue, deeper understanding, and broader consensus.

China's embrace of open-source AI is an important contribution to that effort.

Open models allow researchers and developers far beyond a handful of US technology companies to participate in AI innovation.

They democratize access to knowledge, expand scientific collaboration, and help ensure that AI's future is not dictated by a small group of firms or a single country.

But open access alone is not enough. AI's risks are global, and so must be its governance.

Humanity has cooperated to confront shared threats before in areas ranging from arms control to global security. AI poses a challenge unlike any we have faced, and no nation yet has all the answers.

But we do know one thing: scientists, policymakers, and governments must work together rather than retreat behind geopolitical rivalry.

The future of AI should not be decided by fear or competition alone. It should be shaped by cooperation, shared responsibility, and a commitment to ensuring that this transformative technology serves all of humanity — not just the powerful.

Trust is AI's best accelerator

The rapid pace of AI legislation around the world is an encouraging sign. China is moving decisively in this direction, and AI regulation will mature even faster than the data protection law did. In less than five years, robust legal frameworks could be in place.

Unlike data protection, AI regulation does not exist in isolation. It intersects with other areas of law. The European Union's AI Act alone overlaps with a wide range of legal instruments, creating a regulatory landscape that spans thousands of pages. This is not a criticism — it is simply the reality of governing a technology that touches nearly every sector. At the International Association of Privacy Professionals, we call this digital entropy, where complexity naturally accumulates as technology evolves.

That complexity is already evident. In Ireland, for example, some 15 regulators from different sectors share responsibility for enforcing the EU AI Act, each bringing its own mandate and interpretation. Similarly, within organizations AI governance is divided among teams handling privacy, consumer protection, intellectual property, trust and safety, content moderation and platform accountability. AI now sits at the center of all these functions, making governance inherently cross-disciplinary.

The history of privacy regulation tells us that while global coordination is important, diversity in governance is not a failure. Different countries will inevitably adopt different AI policies, reflecting their own values, cultures and legal traditions. That diversity is a feature, not a flaw. Innovation involves not just designing better technology, but also designing better governance.

Good governance, however, requires investment. Implementing AI responsibly demands people, expertise, funding and technical infrastructure. These are not needs unique to major AI companies. Every bank, insurer and enterprise deploying AI must build its own governance capacity.

As AI reshapes the regulatory landscape, long-standing privacy principles such as notice and consent are becoming harder to apply in practice. Policymakers must therefore look beyond legislation to technical standards and governance frameworks, embedding principles directly into technology.

The Dutch philosopher Desiderius Erasmus once warned that chasing two rabbits means catching neither. In the AI era, that wisdom deserves reconsideration. We should not see innovation and trust as competing goals. Consider the early automobile: before reliable brakes, cars could travel only at modest speeds. It was the invention of brakes — not their absence — that made faster, safer travel possible. Trust and accountability play the same role for AI. They do not constrain innovation; they enable it to move further and faster.

Entropy tells us that systems naturally tend toward disorder, but that tendency is not destiny. With thoughtful governance, strong institutions and responsible technology, we can create more order, more trust and ultimately more innovation in the age of AI.

Putting aspirations into action

Artificial intelligence is a global public good. It must benefit everyone, not just a privileged few. As AI reshapes society, countries must invest in the talent, capabilities and infrastructure needed to participate in the AI era.

Because AI is a global challenge, its governance must also be global. The priority today is not more principles but governance that can be implemented, verified and enforced. Existing AI frameworks remain fragmented, with many relying on voluntary corporate commitments rather than binding rules.

The challenge is not a lack of principles or institutions, but turning them into practical, accountable governance. In this regard, two recent Chinese initiatives stand out: the Global AI Governance Rules Map, and AI Governance Compass developed by the China University of Political Science and Law. The Rules Map offers a real-time, visual overview of AI governance rules worldwide. By breaking down information silos and providing continuously updated information, it improves the visibility, accessibility and comparability of global AI governance. Together with the AI Governance Compass — which systematically presents China's AI policies, regulations and practices — it forms a shared governance infrastructure. These platforms do not impose uniform rules; they enable interoperability. Policymakers can see how other jurisdictions regulate AI, understand why they differ, and identify areas of convergence. In doing so, they turn AI governance from a black box into a transparent, evolving system.

The new AI Governance for Humanity Lab in Valencia, established under the UN Office for Digital and Emerging Technologies, recently released a white paper. It argues that global governance should establish a floor, not a ceiling. A floor sets essential safeguards while allowing countries to adopt stronger protections. A ceiling, by contrast, risks lowering global ambition to the weakest common standard.

Five priorities are particularly urgent. First, establish a globally shared glossary and taxonomy. We cannot govern what we cannot define. Second, make global standard-setting more inclusive by expanding financial support, multilingual resources and regional participation, so developing countries have a stronger voice.

Third, agree on shared guardrails and global red lines to prevent AI from enabling severe human rights abuses or catastrophic risks. Fourth, create a global, machine-readable system for reporting serious AI incidents to strengthen international coordination.

Fifth, align AI governance with climate goals through environmental reporting that keeps AI's growing computational demands consistent with global emissions targets.

China has built a dynamic AI ecosystem spanning research, industry and talent development. Its commitment to openness, multilateral dialogue and international cooperation has contributed valuable experience to global AI governance.

China has also played a constructive role in supporting UNESCO, the United Nations and broader international cooperation.

The future of AI will be built together. Only through cooperation across borders and disciplines can this transformative technology serve all humanity rather than a privileged few.

Ultimately, AI should be judged not by the power of its models, but by the value it creates for people.

Global AI needs global rules

China's rapid advances in open-source AI models have expanded the frontiers of global innovation while prompting fresh debates over data governance, model safety and regulatory standards. Europe, meanwhile, is charting its own course. The European Union has introduced new rules for general-purpose AI and applications that could foster emotional dependency, particularly among minors, drawing clearer regulatory boundaries for the technology. Different jurisdictions are taking different approaches, but they share the same objective: ensuring that AI develops safely, responsibly and sustainably.

The international architecture is beginning to take shape. The UN has convened its first Global Dialogue on AI Governance to promote policy coordination and international cooperation.

It has also established the Independent International Scientific Panel on AI, whose inaugural report — prepared by 40 experts, including two from China — lays the groundwork for a governance model that links scientific assessment with policymaking. The approach mirrors the evolution of climate governance, where science gradually became the foundation of global action.

Yet technology continues to outpace regulation. International consensus cannot be legislated overnight; it must be built through sustained dialogue. The first UN dialogue on AI governance attracted wide international participation.

The next meeting, scheduled for 2027, will be accompanied by a more comprehensive scientific assessment, providing further support for the development of international rules.

Governance, however, is about more than managing risk. It is equally about narrowing the global AI capability gap. China and a group of partner countries are advancing the World AI Cooperation Organization while supporting a UN initiative to strengthen AI capacity, particularly in developing economies.

A global network of regional AI capacity-building centers, launched in Geneva, is already promoting cooperation in talent development, education, knowledge sharing and innovation ecosystems.

The AI revolution also brings new pressures on physical resources. Data centers, electricity demand and water use are emerging as shared global challenges. Future governance must therefore extend beyond algorithms and data to encompass energy, environmental sustainability and resource efficiency.

Participation in UN-led AI initiatives provides an important platform for mutual learning and the sharing of best practices. By learning from one another's experiences, countries can avoid common pitfalls and develop more effective approaches to AI governance.

The views don't necessarily reflect those of China Daily.