Global AI Safety Strategy Depends on U.S.-China Cooperation Amid Mutual Skepticism

As international concern mounts over the potential risks of artificial intelligence, hopes for any unified global approach to safety heavily depend on cooperation between the United States and China. However, both superpowers currently view each other's AI strategies as the primary problem, according to reporting by Chan Ho-Him of ABC News. Artificial intelligence governance is anticipated to be a prominent topic of discussion during an upcoming meeting in Washington between U.S. President Donald Trump and Chinese leader Xi Jinping. The urgency surrounding AI safety has grown in both nations, with American technology leaders emphasizing that averting worst-case scenarios involving rogue systems requires a degree of bilateral cooperation.

Despite the recognized need for coordination, both nations are locked in a race for a decisive technological advantage. President Trump has cautioned that implementing regulations on the technology would ultimately benefit China. This dynamic sets the stage for a presidential meeting where formal, binding agreements may prove difficult to achieve, though initial steps toward broader communication remain possible. Samm Sacks, a senior fellow at the Johns Hopkins School of Advanced International Studies’ Institute for America, China, and the Future of Global Affairs, noted that a formal consensus is nearly impossible at this stage, but suggested that Trump and Xi can create necessary political space simply by acknowledging that AI poses shared risks to both countries.

President Trump has consistently pushed back against domestic calls to strengthen technological regulations, arguing that doing so could allow China to eclipse the United States in the AI sector. Maintaining that the nation leading in technology ultimately wins, Trump stated on social media that the U.S. currently leads China and all others. Meanwhile, the two economic giants are advancing the technology through different strategic approaches. American firms are pushing the boundaries of AI frontiers using immense computing power backed by cutting-edge chips and rapidly expanding data center infrastructure. In contrast, China is pursuing broad global adoption through open-source, more affordable models.

Chinese AI models have achieved significant advancements and gained traction among American corporations, despite U.S.-led export restrictions designed to block China from acquiring the world’s most advanced AI chips. While the majority of American frontier models remain closed-source with closely guarded design elements, U.S. Treasury Secretary Scott Bessent stated that the United States needs to develop more open-source models to effectively counter China. Conversely, the U.S. has accused Beijing of aggressive and malicious attempts to extract capabilities from American AI systems. In response, China has accused Washington of trying to suppress Chinese companies to maintain a monopoly over the global AI industry.

It remains unclear whether the relationship between Trump and Xi can bridge these substantial differences. Nevertheless, figures with ties to the Chinese government view coordination on AI as beneficial. Speaking at the Xiangshan Forum, an annual security and defense conference in Beijing, former Chinese ambassador to the U.S. Cui Tiankai stated that artificial intelligence represents an area where both capitals should actively build up dialogue due to a growing need for cooperation in managing the balance.

Leading American AI companies Anthropic and OpenAI restrict access to their services in China and maintain that protecting America's technological lead is a national security imperative. At the same time, executives from these firms have called for cooperation with Beijing to address safety concerns. In a recent essay advocating for a temporary slowdown in AI development for safety reasons, Anthropic CEO Dario Amodei argued that the U.S. should slow China's progress by maintaining export restrictions on advanced chips. However, Amodei also acknowledged that global pacing will ultimately require cooperation with China, noting it is the autocratic nation with by far the most advanced AI capabilities.

Amodei's remarks drew swift criticism from Beijing. China’s Ministry of Foreign Affairs responded by stating that fearmongering, confrontation, and vicious competition will only disrupt global AI governance and serve no one's interests. The state-run Global Times newspaper published sharper commentary, describing Amodei's remarks as containment provisions rooted in a Cold War playbook for the AI sector. Amid these tensions, Trump has also consulted with Nvidia CEO Jensen Huang, calling the computer chip executive during a podcast interview. Nvidia has historically favored permitting certain chip exports to China to foster an American technological ecosystem and maintain open engagement given the large concentration of researchers in China.

Top officials in both countries have expressed alarm over the increasingly powerful capabilities of AI systems. Chen Yixin, the head of China’s Ministry of State Security, published an article warning that AI poses multifaceted threats ranging from political and ideological security to cyberattacks. Chen argued that advanced systems could threaten China if exploited by actors with ulterior motives, specifically highlighting the powerful models developed by Anthropic and OpenAI as potential vectors for weaponizing cybersecurity vulnerabilities against critical infrastructure. Paul Triolo, a partner at DGA-Albright Stonebridge Group, noted that Chinese concerns have also been stoked by other events, such as a U.S. security firm's recent discovery of AI-related vulnerabilities in WeChat, China's primary messaging and payment platform.

Although Chinese officials indicated following a mid-May summit between Trump and Xi that both nations agreed to pursue dialogue on AI development and governance, little tangible progress has materialized. Lizzi C. Lee, a fellow at the Asia Society Policy Institute’s Center for China Analysis, emphasized that neither country can simply seal itself off from the other. Lee pointed out that both governments share a mutual interest in establishing human-control guardrails, even as they pursue competitive AI strategies for technological advantage.

Simultaneously, both nations have worked to build rival AI alliances with like-minded countries. In July, China launched the World Artificial Intelligence Cooperation Organization alongside 29 initial founding members, including Russia and Pakistan, to promote global governance. Xi has also urged BRICS nations to accelerate the establishment of an open-source AI community and a global governance framework. Political observers widely view the China-led organization as a counterweight to the U.S.-led Pax Silica initiative, which launched last year to strengthen supply chain collaboration on AI with nations such as Japan, the United Kingdom, and Australia.

Looking toward the upcoming summit, Amodei told CBS' "Face the Nation" that a long-term objective would involve working together to establish limits on the pace of AI progress, though he conceded the task will be exceptionally difficult. Ahead of the meetings, a commentary posted by the Chinese state media-linked account Yuyuantantian struck a cautious tone, asserting that substantive discussions can only proceed after the United States first demonstrates that safety rules are applied equally and effectively to its own domestic model companies.