According to Business Insider, more than 1,100 employees from leading AI companies such as OpenAI, Anthropic, Google, and Meta have signed an open letter calling on the U.S. government to slow down the development of AI. This initiative, named "Controlling the Frontier," directly addresses the serious lag in safety and governance systems in the current AI arms race. The open letter focuses on the core issue of "automated AI development"—the ability of future AI to autonomously develop and improve AI systems through "recursive self-improvement." Anthropic has stated that its Claude series models are rapidly approaching this threshold. The letter warns that the development of AI capabilities may accelerate quickly beyond human ability to understand or control these systems, and calls on the U.S. government to support international cooperation to develop necessary technologies and governance tools, consciously controlling the progress of automated AI development.

Executive Backing, Altman's Shift in Stance

The list of signatories of this open letter can be considered an "all-star" lineup in the AI industry—Dario Amodei, CEO of Anthropic, along with two co-founders, Jakub Pachocki, Chief Scientist at OpenAI, Shengjia Zhao, Chief Scientist at Meta, and Anca Dragan, Head of AI Safety and Alignment Work at Google DeepMind, are all included. OpenAI and Anthropic subsequently posted messages on X platform in support; the former acknowledged that at some point in the future, the acceleration of frontier model development might become so fast that the world needs to slow down, while the latter cited its research on recursive self-improvement released last month, emphasizing the need for society to have time to prepare.

Sam Altman, CEO of OpenAI, showed a particularly notable shift in stance. In an interview with the "Invest Like the Best" podcast, he said it might now be time to "slow down" the pace of AI development, allowing society to adapt to new capabilities and establish protective mechanisms.

However, he also emphasized that this approach should not be perceived as regulatory capture or collusion among leading laboratories. Just a year ago, when faced with an open letter advocating a similar "slowing down AI development" proposal, Altman had refused to sign, citing a lack of key technical details. From avoiding signing in 2023 to now publicly supporting it, this change itself reflects a deep transformation in the AI industry's perception of safety risks.