Researchers Say a 10-Year Global Pause on Frontier AI Training Is Feasible

A 26-researcher working group that includes UC Berkeley professors Barry Eichengreen and Stuart Russell says an internationally verified pause on frontier AI training of at least 10 years is feasible. Their 200-page report proposes halting production of AI training chips and replacing them with inference-only chips that cannot be used to train new models. The researchers say success depends on world leaders' cooperation.

• Wesley Holliday, a UC Berkeley philosophy professor, writes that leaders pause if they trust others pause or defections get caught fast.
• Will Fithian, a UC Berkeley statistics professor, writes that bypassing existing AI chip supply chains is extraordinarily difficult.
• Inference-only chips carry hardwired constraints that block frontier training even after theft or seizure, the report says.