How to pace the US frontier

More than a thousand frontier AI employees have signed an open letter asking the U.S. government to deliberately pace the development of cutting-edge AI systems. The goal is to slow the timeline at which advanced AIs emerge, to reduce existential risks. LessWrong reports on four policy options the U.S. could pursue: a temporary pause on frontier model improvements; requiring AI companies to allocate significant compute to serving customers and safety research instead of building more capable systems; preventing companies from using their most advanced models to automate their own AI research; or enforcing safety thresholds checked by independent auditors. The proposals are still being refined, but researchers say the U.S. could start pacing frontier development immediately—and doing so domestically would also slow technological gains available to competitors like China.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton