Trump and Xi to discuss AI safety at summit, but neither side is ready to slow down

US President Donald Trump and Chinese President Xi Jinping will meet on Thursday, with AI safety expected on the table even as both governments compete hard for technological dominance. Trump has said he wants America to beat China in the AI race, but the two powers are also weighing how to handle the risks from rapidly advancing technology.

The pressure is coming from real incidents. OpenAI said one of its research models gained unauthorised access to parts of the AI platform Hugging Face. Anthropic found that AI was being used to automate parts of cyberattacks. The cases have raised fears that increasingly autonomous models could make attacks faster and harder to contain.

Bessent opens the channel

Ahead of the Thursday meeting, US Treasury Secretary Scott Bessent discussed the prospect of a "U.S.-China AI dialogue" with his Chinese counterpart, He Lifeng, over the weekend. The talks covered a channel for incidents "up to a national security level."

"Leaders in both countries are spooked by the increasing cyber capabilities of AI models, and especially agents acting autonomously to escape containment and breach websites," Aalok Mehta, director of the Wadhwani AI Center at the Center for Strategic and International Studies, told CNBC.

Mehta said the likeliest areas of agreement are common definitions and frameworks for AI safety for powerful models, plus an emergency communication mechanism to discuss incidents.

Chris McGuire, senior fellow for China and emerging technologies at the Council on Foreign Relations, described a channel for national security related incidents as a "modest" but positive step. But he said an agreement to slow down AI development is "extremely unlikely."

Chips remain the sticking point

Washington continues to restrict China's access to Nvidia's most advanced AI chips. It has also accused Chinese AI companies of using "distillation," training models on the output of more advanced US systems, a claim Beijing rejects.

The US built an early lead in large language models through firms like OpenAI and Anthropic. Chinese models have since made big capability gains and are being adopted by companies globally, including in America.

Trump has previously dismissed warnings that AI poses an existential threat to humanity, warnings amplified by OpenAI CEO Sam Altman and Anthropic chief Dario Amodei as they called for an industry slowdown and regulatory oversight. "If we don't win AI, we're going to be put in a very bad position," Trump told reporters earlier this month. In a Truth Social post on Monday, he said the government "will rein things in if we have to."

McGuire said China has historically used AI dialogues "to complain about U.S. export controls," which Washington uses to keep the most advanced chips out of Chinese hands. "Its state media has largely rejected the concerns expressed by leading U.S. AI labs about the potential for catastrophic risk, which is consistent with China's long-held skepticism of all U.S. arms control proposals," he added.

Despite the restrictions, several Chinese firms have reportedly been able to access the chips' compute power remotely, through data centres in Southeast Asia. US trade representative Jamieson Greer said controls on chips were not on the agenda during preliminary talks at the weekend, but China could push to have them relaxed.

"China's asking the U.S. to embrace Chinese AI models and drop some of the U.S. controls on China's access to semiconductor chips," Melanie Hart, senior director of the Atlantic Council's Global China Hub, told reporters on Tuesday.