Experts warn Trump's rivalry with China and flawed "AI race" mindset could put the US at risk

A former Huawei executive has pushed back on the framing that a Chinese victory in the race to artificial superintelligence would represent the "greatest AI security threat to the United States," arguing that this narrative is standing in the way of cooperation that could help both nations confront the AI risks actually bearing down on them — dangers to workers, national economies, and security.

Smaller Models Pose the Bigger Danger

Graylin contends that Washington's focus on frontier superintelligence misses where the real threats lie. In his view, the most hazardous systems are frequently compact, specialized models — the kind that rogue non-state actors could build or repurpose to deliver biological, chemical, or cyber weapons. The assumption that the US and China are destined for conflict, he wrote, blocks precisely the global coordination needed to guard against that very tangible danger.

Concrete Steps the US Could Take

Graylin outlined several moves that could ease global anxieties about AI, though some might strike the Trump administration as extreme. Among them: sharing high-level intelligence, opening bilateral and multilateral channels on AI safety, and reviving elements of Joe Biden's "small yard" approach by narrowing export controls. He argued that the current, ever-expanding restrictions are likely preventing China from carrying out more thorough safety testing. If controls were trimmed to target only "genuine military and CBRNE threats" — rather than all advanced technology — China would be better positioned to deploy safer AI products, he suggested.

He also urged Trump to reconsider the administration's deregulatory posture and follow China's example by enacting reasonable safety rules. Doing so, he wrote, could begin to "lessen the growing societal fear over AI that's been brewing domestically," a fear he believes "is certain to hinder the nation's ability to deploy this technology effectively into the economy and reduce our global competitiveness."

What to Watch Next

Graylin's core argument is blunt: "We cannot manage the multifaceted risks of the most powerful technology in human history while refusing to talk to the other nation building it." In an ideal outcome, this week's meeting between Trump and Xi would eventually push Washington to drop the framing of "winning an AI race" altogether. As he told Ars, "There is no way to compete yourself to a safer world."