‘Pacing’ won’t eliminate the risk of AI doom. Here’s what could | David Krueger

Summary

David Krueger argues that Jacob Coxon’s resignation from Anthropic demonstrates a serious turning point in public awareness of artificial-intelligence risks. He criticizes Anthropic CEO Dario Amodei’s proposal to “pace the frontier” through a one- to two-year slowdown and technical-safety research, saying it does not adequately address uncertainty about how advanced AI systems work, how they might evade safety evaluations, or whether humans can remain in control of systems more capable than people. Krueger instead calls for an immediate, indefinite international moratorium on frontier-AI development. He proposes decommissioning advanced data centers and chips in the United States and China, potentially followed by dismantling parts of the advanced-chip supply chain, arguing that this would be easier to verify than restrictions on chip use. He acknowledges substantial economic costs and says he is researching less disruptive alternatives. The article also criticizes external-evaluation proposals as expanded industry self-regulation and says current AI governance is effectively “safe-until-proven dangerous.” Krueger urges cooperation among the United States, China, and other governments to halt the AI race. The piece is an advocacy argument by a University of Montreal professor, not a report establishing that the described policies have been adopted or that extinction is imminent.