President Trump is traveling to China this week to discuss artificial intelligence safeguards with President Xi Jinping, marking the first serious attempt by both superpowers to establish ground rules for deploying increasingly powerful AI systems that could be weaponized in cyberattacks.
The meeting reflects a paradox that now defines great power competition: both Washington and Beijing recognize they need shared protocols to prevent AI-enabled attacks, yet both are simultaneously racing to build AI capabilities for offensive cyber operations and to slow the other's technological progress.
U.S. officials confirmed Sunday that Trump plans to open a dialogue on AI security norms during the Beijing visit. "We want to take this opportunity with the leaders meeting to open up a conversation and see if we should establish a channel of communication on AI matters," one official said.
The U.S. has deployed export controls to restrict China's access to advanced semiconductors needed for AI development, but officials increasingly acknowledge that containment alone cannot address the security risks posed by powerful AI systems that both nations will inevitably possess.
Sixteen business executives are joining Trump on the trip, though notably absent are leaders from major AI firms. The exclusion is striking given that American AI companies are grappling with how to release increasingly sophisticated models without creating tools that hackers and foreign intelligence agencies could exploit to find and weaponize software vulnerabilities.
The White House spent weeks debating how to regulate the release of powerful AI models after months of resisting regulation entirely. Meanwhile, it accused China last month of conducting "industrial-scale" campaigns to steal and replicate American AI models.
Both countries have already begun testing offensive cyber capabilities powered by frontier AI systems. Anthropic, an AI safety firm, disclosed in November that Beijing used its Claude model to automate parts of a spy campaign targeting roughly 30 organizations worldwide. The U.S. National Security Agency, which runs major espionage operations, is testing AI systems from Mythos.
That mutual engagement in AI-powered cyber operations creates a credibility problem for any pledge of restraint. Melanie Hart, a China expert at the Atlantic Council and former State Department official, said Beijing used previous AI safety discussions under the Biden administration primarily to extract U.S. intelligence rather than negotiate genuine guardrails.
"The topic is important enough and dangerous enough that we should be having engagement with China on this," Hart told reporters. She noted, however, that Chinese delegations at those earlier talks often included foreign ministry representatives without technical AI expertise, suggesting a lack of serious intent.
Hart cautioned against expecting a single visit to reshape U.S. policy. The trip's real value, she said, will be determined by whether future talks produce substantive agreements or remain largely symbolic. "From there, we then need to judge who shows up for the China side," Hart said. "We want to see the technical experts showing up at the table. That's how we'll know that that's actually real."
Author James Rodriguez: "Without both sides sending actual technologists to the table, this whole exercise is just expensive theater masking a competition that neither country is willing to genuinely constrain."
Comments