Anthropic's CEO Calls for an AI Slowdown, Markets Tumble

Dario Amodei picked an unusual moment to warn the AI industry against its own speed. In an essay titled "We Must Pace the Frontier," the Anthropic chief executive argued that companies are building recklessly, warning that autonomous AI agents could eventually trigger hundreds of billions of dollars in damage if left unchecked.

The call landed hard on markets already jittery about AI valuations. SoftBank shares slumped in Tokyo, with the Guardian reporting a drop of roughly 13% while other accounts put the fall closer to 11%, a gap that reflects how fast the sell-off moved during the session rather than any dispute over direction.

South Korea's KOSPI index, heavily weighted toward chipmakers that supply the AI buildout, fell somewhere between 3% and 3.7% depending on the source. SK Hynix dropped 5.75%, and Taiwan Semiconductor Manufacturing Company slid 1.2% in Taipei. Nasdaq futures pointed to a US tech selloff at the open, with estimates of the drop ranging from 1.3% to 1.8%.

What makes the moment notable is who agreed with Amodei. Sam Altman, Demis Hassabis and Elon Musk all voiced public support for pacing AI development, an unusual moment of alignment among rivals who normally compete to ship faster. Altman went further, committing OpenAI to embedding outside evaluators to audit its safety practices, a pledge that mirrors one Amodei had already made for Anthropic.

Not everyone in the room agreed. Donald Trump dismissed the idea of slowing down outright.

"We're leading China in AI. We're the most sophisticated country in the world and frankly, I want to keep it that way because whoever wins AI, wins."

That tension, between an industry now openly questioning its own pace and a political leadership treating AI dominance as a zero-sum race with China, is the real story behind the market swings.

Anthropic's position carries some weight because the company says it isn't just posturing. It told investors it's on track for a second consecutive profitable quarter, meaning the slowdown message doesn't come from a company scrambling to justify losses.

Separately, Altman confirmed OpenAI will not pursue an IPO this year, removing one near-term catalyst investors had been watching. Meanwhile, a cross-party group of UK MPs and peers added a policy dimension to the same news cycle, warning that no country currently has laws adequate to contain the human rights risks AI systems pose.

Taken together, the day's events suggest AI's biggest builders are starting to publicly wrestle with the pace they set, even as markets treat any hint of deceleration as a reason to sell chip and AI-adjacent stocks first and ask questions later.