Anthropic's call for AI development slowdown

Anthropic CEO Dario Amodei published an essay, "We Must Pace the Frontier," on September 12 calling for AI companies to voluntarily slow the pace of capability advances to give safety research time to keep up, warning that without a slowdown, AI could within six to 12 months be capable of directing swarms of agents that could "take over the entire internet." He proposed independent, embedded evaluators with ongoing access to frontier AI labs, industry-wide safety standards, and continued US restrictions on advanced AI chip sales to China. OpenAI's Sam Altman and Elon Musk quickly voiced agreement, and OpenAI said it would commit to embedding outside evaluators; Altman also said OpenAI would delay its stock market debut until at least 2027 to focus on safety. The essay followed the high-profile resignations of two Anthropic safety researchers, Jacob Coxon and Joe Benton, who said AI firms were "gambling with our lives" in a race toward self-improving superintelligence.

Reaction was split: some Silicon Valley figures, including Nvidia's Jensen Huang and investors at a San Francisco tech conference, dismissed the warnings as hype ahead of Anthropic's and OpenAI's expected IPOs, while Sen. Bernie Sanders cited the warnings in pushing his Ban Artificial Superintelligence Act. President Trump said he had no concerns about AI causing human extinction and reiterated the US needed to beat China in AI; China's foreign ministry accused Amodei of "fearmongering" and using safety rhetoric to justify chip restrictions aimed at containing China.