AI Safety Researcher David Krueger Calls for Frontier AI Moratorium

AI safety researcher David Krueger called for an immediate, indefinite international moratorium on frontier AI development in a September 10 social media post. Krueger, who has studied AI alignment for over a decade, argued that current regulatory frameworks are inadequate because existing techniques cannot reliably verify system safety or control. He stated that any effective safety regulation must inherently include a development pause, as current methods fail to guarantee safe operation. The proposal advocates for halting frontier model training until verifiable control mechanisms exist. Krueger’s position represents a significant escalation from standard reformist safety advocacy to explicit cessationism. His argument rests on the technical claim that reliability gaps in safety evaluation make continued scaling inherently dangerous regardless of policy guardrails. This stance challenges industry assumptions that iterative deployment and regulation can manage risks during ongoing capability advancement.