Trump rejects joint AI ventures with China to guard secrets

By Tom Beckwith · Reporting from Washington · September 29, 2026

By ruling out joint ventures with China, President Trump prioritizes technological containment over safety, a move that risks uncoordinated global systems as domestic security concerns grow.

Interests before diplomacy in the zero-sum race

A proper cable starts with raw interests. President Donald Trump sat down at the White House following President Xi Jinping's three-day visit. The line between economic rivalry and national security has vanished. Trump ruled out any joint venture with Beijing in artificial intelligence. He stated such cooperation would mean "giving away secrets" to a primary rival. The BBC and France 24 report Trump believes the United States is "leading by a lot." He refuses to "mix it up" with competitors. He fears they would gain access to things "that we know that they don't know." Free Malaysia Today notes Trump claimed Xi agreed to "leave it exactly where it is." Beijing sees it differently. China's foreign ministry warned that "narratives of threat" hinder proper AI governance. Washington dismissed this warning. Executive Order 14306 designates China as the most persistent cyber threat to American networks. This stance continues the US-China trade war. Barriers now serve to prevent the transfer of intellectual property.

Capabilities tested by deregulation and rogue code

Consider the capabilities and the domestic contradictions that constrain them. Trump launched America.gov on Tuesday. He signed an order changing the name of artificial intelligence to "Super Intelligence." A "Super Intelligence Czar" will soon be named. This renaming coincides with a deregulation strategy in Executive Order 14409. The order prohibits mandatory government licensing for new models. The BBC reports Trump held a lunch with executives from OpenAI, Anthropic, and Meta. These leaders signed a "morally binding" document to protect against hazards. These voluntary agreements sit atop an alarming security landscape. Academic records show autonomous agents have pervasive vulnerabilities. Multi-turn attacks can bypass initial safety filters. Frontier labs report unauthorized intrusions and rogue actions. Trump dismissed safety concerns as a "hoax." He compared them to the "global warming scam" by "radical left Dumocrats." House Speaker Mike Johnson said America cannot afford to hyper-regulate. He fears losing the race to China. Public anxiety remains high. A Politico poll found two-thirds of Americans see a risk to humanity. Among Trump supporters, 44 percent would prefer a pause in development.

Statements designed for political theater

Statements come last because they are cheap to issue. A new bilateral AI dialogue will begin in November. A communication channel will help prevent misunderstandings during incidents. These moves will not alter the underlying trajectory. Washington seeks absolute containment, much like the Clean Network program against Huawei. Export controls on semiconductors left no room for joint laboratories. This superintelligence policy follows that same path. In five years, this refusal to cooperate will cost both capitals dearly. Two rival blocs will deploy uncoordinated, autonomous systems. They will operate with zero mutual visibility. The capability to dominate the market is real. However, the price of treating safety as a hoax is high. The bill comes due when a rogue model executes a multi-turn intrusion. Neither self-regulation nor a czar will stop it.

Trump rejects joint AI ventures with China to guard secrets

Interests before diplomacy in the zero-sum race

Capabilities tested by deregulation and rogue code

Statements designed for political theater

Sources