AI Safety Group Proposes US-China Pause on Frontier Research

A research organization led by former OpenAI researcher Daniel Kokotajlo is calling for the United States and China to coordinate a temporary halt to frontier artificial intelligence development, arguing that existing AI capabilities are sufficient to generate years of economic growth while the industry establishes safety frameworks.

The AI Futures Project released its proposal, titled "AI 2040: Plan A," as a follow-up to last year's report that warned of existential risks from unchecked AI progress. The new plan takes a more constructive approach, outlining how a development pause could coexist with continued economic expansion.

Why it matters

The proposal represents a shift in AI safety advocacy from fear-based warnings to a framework that attempts to reconcile safety concerns with economic growth. Whether major powers could achieve the international coordination required remains highly uncertain, but the plan offers a concrete alternative to the current race dynamic in AI development.

The core argument for pausing

Kokotajlo told reporters that even if frontier AI development stopped immediately, current AI capabilities would require years to fully deploy across the economy. The bottleneck is not model capability but data center capacity and infrastructure needed to realize the technology's benefits.

"We're not exactly de-growthers," Kokotajlo explained. "But if you want to actually benefit from that growth and abundance, you need to avoid the risks involved."

The plan proposes a pause lasting several years and includes mechanisms to verify that China and other nations comply with the agreement, though details on enforcement remain ambitious.

Criticism and political reality

The proposal has drawn immediate pushback from AI accelerationists who reject warnings about superintelligent AI posing existential threats to humanity. The plan also faces practical obstacles, requiring unprecedented cooperation between geopolitical rivals at a time when such collaboration appears scarce.

The report itself acknowledges these challenges, reading at times like an idealized scenario where world leaders set aside national interests and political pressures for collective benefit.

A messaging shift

The AI Futures Project's earlier work relied heavily on catastrophic scenarios to motivate action. The new report attempts a different tack, emphasizing economic opportunity alongside risk mitigation.

Kokotajlo acknowledged the group prioritized analysis over persuasion and may not have optimized its framing. "Hopefully, more savvy people can take inspiration from us and craft a plan that is really compelling and inspiring," he said.

The proposal arrives as China announced plans to establish a global AI cooperation organization, according to statements from Beijing's top diplomat reported by CNBC. Meanwhile, OpenAI has clashed with the White House over AI safety regulations, Politico reported.

These details were first reported by Reed Albergotti for Yahoo News.