Two world powers are in an arms race to develop the most advanced AI systems, and neither of them trusts each other—but each relies on the other’s compliance to proceed. This contradiction lies at the core of a dangerous standoff for our time.
President Trump’s meetings with President Xi Jinping in Beijing this week are a crucial moment for the U.S.-China relationship. U.S. officials made clear their intentions to initiate the discussion of setting up a dedicated communication channel regarding AI matters This means they’re worried that the technology could become a source of conflict between the two nations.
I’ve been working in the tech world for decades, and I’m confident that this situation is unprecedented. Washington and Beijing both see the importance of advanced AI technology for purposes of intelligence and as a potential means of cyber warfare. It has thus become extremely important to coordinate and cooperate, even while remaining rivals.
The U.S. has relied significantly on export control of technologies and equipment to impede AI development in China. At the same time, it has become increasingly obvious that blocking China from importing chips alone does not solve the problem. Even if you slow your rivals’ AI development, there will remain a scenario where both sides employ it in the context of offense without any set regulations.
Chinese AI models such as DeepSeek compete on the global market as worthy contenders to American products. Additionally, according to recent accusations by the White House, Beijing launched industrial-scale operations aimed at extracting and copying American AI models. The irony is that both nations have experimented with using AI as an instrument of offensive cyber attacks.
It’s become evident that the U.S. and China are simultaneously developing offensive tactics based on AI models, which makes any call for restraint in such matters hypocritical. Yet that is quite logical under current conditions. In a security dilemma, it is difficult to trust your adversary.
Domestic issues complicate this matter
There are also internal problems. The American companies working with AI find themselves at odds with U.S. regulators, who have not come up with reasonable guidelines for releasing new models. The discussions have been ongoing for several months now, and American companies have been opposed to government regulation for years.
This lack of clarity on domestic grounds weakens American positions during negotiations with Chinese authorities.
According to Melanie Hart, a former official with the State Department working for the Atlantic Council, AI is too significant to leave China out of the discussion. However, it should be mentioned that previously, Beijing officials used AI security discussions held under the Biden administration to gather information rather than discuss possible restrictions. They even employed foreign ministry representatives who lacked technical AI knowledge.
This information shows that there might be some reason for suspicion, but does not warrant stopping all negotiations.
While the summit cannot become the event where U.S. AI policy takes a revolutionary turn, it can help figure out whether further discussions on the safety of AI technology will be substantial or merely ceremonial.
That distinction has become important because of rapid technological advancements. Advanced AI that can detect software vulnerabilities has become a threat to everyone. Systems of this sort cannot be controlled by governments yet. The situation is getting out of hand.
The AI arms race is a reality, and it’s impossible to stop it at this point. However, the problem here is whether it is possible to be rivals and partners simultaneously and compete fiercely while still being able to talk to each other about this critical issue.