Skip to main content

1100 AI insiders to Washington: Hit the brakes on AI

Over a thousand employees from the world's leading AI companies are telling Washington: slow down. In an open letter called "Controlling the Frontier," more than 1,100 workers from OpenAI, Anthropic, Google, and Meta have joined forces to call for a deliberate pause in the race to build ever-more-powerful AI.

The letter zeroes in on a specific worry: "automated AI development." That's the idea that future AI systems could design and improve other AI systems on their own, without human help—a process known as recursive self-improvement. Anthropic has already said its Claude models are getting close to that threshold. The signatories warn that once AI starts improving itself, things could accelerate faster than we can understand or control.

An all-star lineup

The list of signatories reads like a who's who of AI. It includes Dario Amodei, CEO of Anthropic, along with two of its co-founders; Jakub Pachocki, chief scientist at OpenAI; Shengjia Zhao, chief scientist at Meta; and Anca Dragan, who leads AI safety and alignment at Google DeepMind. Both OpenAI and Anthropic posted messages on X supporting the letter. OpenAI acknowledged that at some point, frontier model development might speed up so much that the world needs to slow down. Anthropic pointed to its own research on recursive self-improvement, saying society needs time to prepare.

Altman's about-face

Perhaps the most striking shift comes from Sam Altman, CEO of OpenAI. In a recent interview on the "Invest Like the Best" podcast, he said it might now be time to "slow down" AI development so society can adapt and build safeguards. But he also stressed that this shouldn't be seen as regulatory capture or collusion among big labs.

Just a year ago, Altman refused to sign a similar open letter calling for a slowdown, citing a lack of technical details. Going from refusing to sign in 2023 to publicly backing the idea in 2026 shows just how much the industry's view of safety risks has changed.

What happens next?

The letter calls on the U.S. government to support international cooperation and develop the tools needed to govern automated AI development. It's a direct challenge to the current arms-race mentality, where companies compete to release the most powerful models as fast as possible.

Whether Washington will listen remains to be seen. But with more than a thousand insiders—including some of the biggest names in the field—saying it's time to slow down, the message is hard to ignore.

Key Points

  • Over 1,100 employees from OpenAI, Anthropic, Google, and Meta signed an open letter urging the U.S. government to slow AI development.
  • The letter focuses on "automated AI development"—AI that can improve itself recursively—which could accelerate beyond human control.
  • Signatories include top executives and scientists like Dario Amodei, Jakub Pachocki, and Anca Dragan.
  • Sam Altman, who previously refused to sign a similar letter, now supports a slowdown.
  • The letter calls for international cooperation and governance tools to manage the pace of AI progress.