U.S.-China AI Cooperation Faces Mutual Distrust Before Summit

Growing urgency, deepening divide

Artificial intelligence governance will be on the agenda when President Donald Trump meets Chinese leader Xi Jinping in Washington next week, according to the Associated Press. The meeting comes as both superpowers grow increasingly wary of each other's AI ambitions, even as American technology leaders warn that managing AI risks requires some level of cooperation with China.

The challenge is stark: both nations are racing for AI dominance while viewing the other as a threat. Trump has repeatedly argued that regulating AI would hand China an advantage, declaring that whoever leads in AI "wins." Meanwhile, China accuses the U.S. of attempting to monopolize the industry through export restrictions on advanced chips.

Why it matters

Without coordination between the world's two AI superpowers, establishing global safety standards or preventing worst-case scenarios becomes nearly impossible. Yet the current trajectory suggests both countries prioritize competitive advantage over collaborative risk management, potentially leaving critical AI governance gaps that affect all nations.

Different paths to AI leadership

The two countries are advancing AI through contrasting strategies. U.S. companies leverage cutting-edge chips and expanding data center infrastructure to push AI frontiers, while China pursues global adoption of more affordable, open-source models.

Chinese AI models have made significant progress despite U.S.-led restrictions blocking China's access to the most advanced AI chips. U.S. Treasury Secretary Scott Bessent said Tuesday the U.S. needs to develop more open-source models to counter China's approach, even though most American frontier models remain closed-source.

The U.S. has accused China of "aggressive, malicious" efforts to extract capabilities from American AI systems. China counters that the U.S. seeks to suppress Chinese companies to build an AI monopoly.

Calls for cooperation meet political reality

Anthropic CEO Dario Amodei wrote in an essay last weekend that "global pacing will require cooperation with China, the autocratic country with by far the most advanced AI capabilities." He simultaneously called for maintaining U.S. chip export restrictions to slow China's progress.

His remarks drew sharp criticism from Beijing. China's Foreign Ministry said Monday that "fearmongering, confrontation and vicious competition will only disrupt the process of global AI governance." State-run Global Times called his essay a "Cold War playbook" for AI.

Both OpenAI and Anthropic restrict access in China while advocating for U.S. technological leadership alongside safety cooperation.

Security concerns drive both sides

Chen Yixin, head of China's Ministry of State Security, warned Sunday that AI poses threats to the country's political security and critical infrastructure. He specifically cited Anthropic's and OpenAI's models as potential risks.

Recent discoveries of AI vulnerabilities in WeChat, China's dominant messaging and payment platform, have heightened Chinese concerns about AI governance, according to Paul Triolo of DGA-Albright Stonebridge Group.

Limited expectations for the summit

Experts anticipate any agreements will likely be superficial. "A formal agreement or consensus may be near impossible, but they do not need to go that far," said Samm Sacks of Johns Hopkins School of Advanced International Studies. "Trump and Xi just need to create political space by acknowledging AI poses risks to both countries."

While Chinese officials said after a mid-May summit that the two countries agreed to pursue AI dialogue, little progress has materialized. Both nations have instead focused on building AI alliances with aligned countries—China through its World Artificial Intelligence Cooperation Organization launched in July, and the U.S. through the Pax Silica initiative.

Lizzi C. Lee of the Asia Society Policy Institute noted that neither government will abandon its competitive AI strategy, but both have mutual interests in establishing guardrails for human control of AI.

These details were first reported by the Associated Press.