Pacing AI development won't prevent catastrophic risk, says safety researcher

David Krueger, a safety researcher, argues that slowing AI development by one to two years—a proposal endorsed by Anthropic CEO Dario Amodei, Sam Altman, and Elon Musk—does not adequately address the risks of advanced AI systems. According to Krueger, experts have long understood that AI companies lack adequate control mechanisms for these systems, and pacing alone cannot eliminate that fundamental vulnerability.

Krueger's critique follows Jacob Coxon's resignation from Anthropic, which Krueger characterizes as marking a tipping point in public awareness of AI safety concerns. He contends that the industry needs a concrete plan to ensure AI systems remain controllable, rather than a proposed slowdown in development pace.