Van Hollen Urges Trump to Negotiate AI Guardrails With Xi

Senator Chris Van Hollen led a group of 16 Senate Democrats on Tuesday in urging President Donald Trump to use his planned summit with Chinese President Xi Jinping to negotiate a formal agreement on guardrails for frontier artificial intelligence, including a possible mutual pause in the development of the most advanced AI systems.

The letter, dated September 23, calls on Trump to engage Xi directly on AI guardrails and global standards for how frontier AI models are built, tested, and used. The lawmakers argue that a US-China agreement on AI safety would set the standard for how the technology is used around the world.

The senators also asked Trump to consider slowing the pace of frontier AI development, or agreeing to a mutual pause, to guard against cyberattacks, the development of biological weapons, and other dangerous uses of the technology, according to the letter.

The signatories include Senators Patty Murray, Elizabeth Warren, Tim Kaine, Maria Cantwell, Richard Blumenthal, Mazie Hirono, Tammy Duckworth, Catherine Cortez Masto, Tina Smith, Jacky Rosen, Ben Ray Luján, Alex Padilla, Raphael Warnock, Peter Welch, and Angela Alsobrooks, joining Van Hollen, a Maryland Democrat, in pressing the White House ahead of the meeting.

What the senators want from the summit

The letter asks for discussions at the highest level covering the standards for development and testing of AI models, oversight and monitoring of those models, rules to keep human control over autonomous systems, and ways to technically verify that any agreement is being honored.

Van Hollen and his colleagues pointed to safety incidents as evidence that current oversight is not enough. Testing AI models without constant monitoring and safe sandboxing has produced security breaks, the letter says, and publicly available models are already capable of finding and exploiting cybersecurity vulnerabilities. Companies building AI have even more powerful models internally, according to the letter.

The senators also criticized the administration for loosening semiconductor export controls on China and refusing to close loopholes in the remaining regulations, arguing that the policy lets Beijing access the US-made chips it needs to speed up its own AI programs. Both countries are now integrating AI tools into military and defense applications, including autonomous and unmanned systems, which the letter warns against without significant safeguards.

Why they are raising the alarm now

The letter leans on warnings from inside and outside government. CIA Director John Ratcliffe recently said it would "not be misplaced" to describe AI capabilities as "akin to digital nuclear weapons," according to the letter. In July, more than 1,300 employees of frontier AI companies called on the US government to work internationally on tools to pace the development of AI, particularly as automation of AI research itself accelerates progress.

Former Treasury Secretaries Henry Paulson and Robert Rubin recently published an opinion piece calling on Trump and Xi to work toward what they described as an AI Cooperation Treaty, and the senators cite that proposal in pressing for the summit to produce something concrete.

Both Washington and Beijing have already put some rules in place, the letter acknowledges. The Trump administration implemented a voluntary testing program before the public release of advanced AI models, and China has adopted registration and safety requirements for certain applications, though the senators note that consistent enforcement remains uncertain. They argue that neither side's domestic rules are sufficient on their own.

The senators draw a parallel with Cold War diplomacy. The United States and the Soviet Union negotiated multiple nuclear arms control agreements despite deep ideological differences, they write, and verification measures sat at the core of those deals. Crafting the technical specifications for AI guardrails, they argue, will take the same combination of expertise and diplomacy.

The summit is expected to take place on September 24, and public reporting indicates AI is already expected to come up in the discussions between the two leaders, according to the letter. The full text of the letter is available on Senator Van Hollen's website.