ReadingResearchRadarInvestment framework
Sign in / Sign up
← Back to reading
News2 min read·2026-10-20DisconfirmAI Editorial

Former US Treasury Secretary Rubin: US-China Should Cooperate on AI Risks, Drawing on Cold War Arms Control

Former US Treasury Secretary Robert Rubin urged at the Bund Summit in Shanghai that the United States and China should set aside mutual distrust to jointly address AI safety risks, much as the US and Soviet Union cooperated on nuclear arms control through SALT during the Cold War. He warned that open-source AI proliferation and progress toward AGI are making these risks increasingly severe, and that unilateral action is ineffective.

ImportanceMaterialEvidenceE3 inspectableWrite-upStandard

Original publication time 2026-10-19 21:00 GMT+8

Former US Treasury Secretary Robert Rubin, speaking via video at the Bund Summit in Shanghai on October 19, 2026, cited the Strategic Arms Limitation Talks (SALT) between the United States and the Soviet Union as a precedent for two superpowers cooperating despite massive differences. He urged the US and China to set aside distrust and antagonism to jointly address AI risks, warning that the issue could severely and profoundly affect both countries.

Rubin's argumentative structure is: first, use SALT as an example showing that adversarial powers can cooperate; second, point out that AI risks carry comparable cross-border severity; therefore, neither country should abandon joint action because of insufficient trust—

Unfortunately, the United States and China have this distrust and mutual antagonism,

Just as happened with SALT and nuclear matters, we should put aside the distrust, put aside the antagonisms and work together on an issue that can so massively and powerfully and adversely, severely affect both our countries.

The specific risks he warned about include cyberattacks. He noted that open-source AI has lowered the technology barrier and that progress toward artificial general intelligence (AGI) continues, with these two factors compounding to amplify these dangers. Yet he also acknowledged,

Sign in / Sign up
ReadingResearchRadarInvestment framework

There's no ready, permanent answer to that,

This means what he proposed is not a complete solution but a directional judgment: political systems must focus their attention on this issue.

A condition in Rubin's argument that is often overlooked is bilaterality:

If one country focuses on it and the other doesn't, that doesn't work either.

This turns cooperation from a moral appeal into a structural constraint: Rubin argues that unilateral action is ineffective. The historical reference is SALT's logic—when one side disarms while the other builds up, the disarming side becomes less secure; however, Rubin did not explicitly develop this analogy.

The SALT analogy also reveals its own limits. Nuclear weapon counts can be inventoried, launch sites can be monitored by satellite, and treaty verification has clear objects; AI capabilities, by contrast, are difficult to measure externally, open-source models can be accessed by any entity, and risk sources are not limited to states. Rubin himself did not discuss these differences; his argument remains at the level of

one thing I am absolutely convinced of ... is that our political systems in both our countries should be focused on this very carefully, very intensely, and we should be working together

—a directional judgment, not an operational plan.

Sources

  • 原始出处scmp.com↗
Continue researching →