On Wednesday, Donald Trump will become the first sitting US president to visit China in nearly a decade. The agenda is packed: the Iran war, trade tensions, Taiwan, nuclear proliferation. But behind the expected diplomatic theater, there is one issue where Washington and Beijing might actually find common ground.

AI.

Both governments are scared. Genuinely, sincerely scared. And the catalyst has a name: Mythos.

When Anthropic flagged its own model to the Trump administration as a potential national security concern, it triggered a pivot that nobody in Washington saw coming. The White House, which had revoked Biden's AI safety executive order on day one and spent months championing a hands-off approach, suddenly found itself scrambling for guardrails. White House Chief of Staff Susie Wiles personally met with Anthropic CEO Dario Amodei. The intelligence community started demanding pre-release model vetting. And now, heading into Beijing, the administration appears poised to reshape its entire approach to AI security.

Axios reported that Iran, China, and AI are colliding in what it called Trump's "legacy-defining week." The White House's evolving posture, according to Axios, "marks a pivot from its earlier laissez-faire approach to AI, driven by fears that the technology is advancing faster than governments can control it."

On the Chinese side, the picture is almost the mirror opposite. Beijing has regulated AI from the start. China imposed mandatory safety assessments for generative AI models in 2023. It required watermarking of AI-generated content before anyone in the West even proposed it. It has an AI classification system that makes the EU's risk-based approach look casual.

But there is a fascinating gap. CNBC's Evelyn Cheng, reporting from Beijing, highlighted a striking cultural difference: Chinese students and businesses are enthusiastically embracing AI, while the American public increasingly worries about its negative impact. Pew Research found that only 10% of Americans are excited about AI. In China, the technology is woven into daily life from shopping to education to healthcare.

This creates a peculiar dynamic for the summit. Trump needs to show he is serious about AI safety after months of deregulation. Xi needs to show he is open to international cooperation while maintaining domestic AI supremacy. Both need to address the chip export controls that have become the primary friction point in the tech relationship.

The South China Morning Post framed it as a stark choice: escalate or relax chip controls. CSIS, the think tank, expects the summit will focus on economy and Iran from the US side, while China will push for stability on Taiwan and progress on trade. But the LA Times reported that "fears of an AI breakthrough" are what ultimately forced both sides to the table on technology cooperation.

We covered the US-China emergency AI talks in our earlier reporting. Those were preliminary. This is the main event. The question is whether two governments that cannot agree on tariffs, territories, or a war in the Middle East can somehow agree on rules for technology that neither fully understands.

If history is any guide, they will issue a vague joint statement about "responsible AI development," shake hands for the cameras, and return to competing as aggressively as ever. But even a vague statement would be significant. It would be the first time the US and China formally acknowledged AI as a bilateral governance challenge rather than purely a competitive one.

The summit runs May 14 to 15. Watch for language on AI in whatever communique emerges. The details will be in the words they choose, and the commitments they avoid.

Sources: Axios, CNBC, Reuters, New York Times, LA Times, South China Morning Post, CSIS.