← All stories

People and power

Anthropic's CEO asks AI to slow down, rivals agree, and tech stocks wobble

On September 12 Dario Amodei called for slowing the most advanced AI, the bosses of OpenAI, xAI and Microsoft voiced support, and tech shares dipped the following Monday.

It is not every day that the head of a company asks his whole industry, his own firm included, to ease off the accelerator. On September 12, 2026, Dario Amodei, chief executive of Anthropic (the company behind the Claude chatbot), did exactly that. Within a day, the bosses of several of his biggest competitors said they agreed. By Monday, tech stocks had dipped.

What the essay asked for

Amodei published the essay, We Must Pace the Frontier, on his personal website. In AI, the frontier means the most advanced models: the ones pushing the limits of what AI can do. As summarized by the tech news site MarkTechPost, his argument is that the industry should slow down how quickly those models gain new abilities, so that safety work can catch up.

The essay sets out three steps:

  1. Outside inspectors inside the labs. Independent evaluators would get employee-like access to AI companies to check their work.
  2. Shared rules among US labs. Common safety standards, plus agreed checkpoints for what new models are allowed to do.
  3. Talks with authoritarian countries about limits on the riskiest AI.

Anthropic committed to doing only the first step on its own. The other two depend on other companies and governments joining in.

In an interview with CBS Sunday Morning, Amodei went further than most executives ever do. He told CBS that "for too long the industry lied to people" about AI's risks. He compared the evaluator idea to a food inspector for AI. CBS also reported that he supports federal regulation and possible emergency shut-off switches, but opposes outright bans.

Rivals said yes, mostly

The real surprise was the reaction. According to MarkTechPost, OpenAI's Sam Altman wrote the same day that embedding evaluators was a great idea and that OpenAI would do the same. Elon Musk, whose company xAI makes the Grok chatbot, posted that Amodei was right. On September 13, Microsoft's Satya Nadella welcomed a more deliberate pace, with a catch: he said any such system must not end up controlled by a small handful of players. A later lawsuit quotes Google DeepMind's Demis Hassabis praising the essay as the right way ahead, Yahoo Finance reported.

Not everyone was impressed. A Bloomberg Opinion column argued that the warning did not go nearly far enough. And rivals agreeing in public turned out to carry a legal risk of its own, which we cover in a separate post about an antitrust lawsuit.

A wobble on the stock market

On Monday September 14, tech shares fell. CBS News reported that on that day the Nasdaq Composite, a stock index packed with tech companies, dropped 0.6% to 26,186. The broader S&P 500 slipped 0.5% and the Dow 0.3%. Chipmaker Nvidia fell 3.4% and memory-chip maker Micron 5.3%. Chip shares were hit hardest: the Philadelphia Semiconductor Index, which tracks 30 big chip companies, fell 5.9%, with every one of them down, according to the market-news site TradingKey.

Two words of caution. First, for the market as a whole these were small moves: a wobble, not a crash, even if chip shares took a real knock. Second, news coverage linked the dip to the essay, but nobody has shown that the essay caused it. Markets react to many things at once. Nothing here is a reason to buy or sell anything.

What we don't know yet

Sources

Keep reading

New to AI? Start with which assistant to use, how to stop AI training on your chats, which ChatGPT plan to pay for or how to write a good prompt.