In a comment piece, AI researcher David Krueger argues that Anthropic chief executive Dario Amodei's proposal to "pace the frontier" of AI development – a one-to-two-year slowdown to allow "profound progress" on safety – is inadequate to address the existential risks posed by advanced AI. Krueger, an AI professor at Cambridge and Montreal, contends that fundamental problems remain unsolved: researchers do not understand how these systems work, cannot reliably prevent them misbehaving, and have no way of staying in control once AI surpasses human intelligence, meaning a short pause cannot realistically fix these issues.

Instead, Krueger calls for an immediate, indefinite, international moratorium on frontier AI development, proposing that the US and China jointly decommission advanced computer chips and datacentres, and dismantle the concentrated supply chain for chip-making technology such as extreme and deep ultraviolet lithography, to make any covert breach easily detectable by intelligence services. He acknowledges this would carry substantial economic costs, comparing it to pandemic-era shutdowns, but argues these are justified given the stakes, and criticises Amodei's plan for ultimately still pursuing "recursive self-improvement" of AI with diminishing human oversight.