Over 1,171 scientists and employees from major AI companies have issued a formal statement requesting U.S. government support for efforts to intentionally slow the pace of artificial intelligence development. The signatories include Dario Amodei of Anthropic, Jacob Paciocchi of OpenAI, and Zhao Shengjia of Meta AI, among others, who warn that automation of AI research itself is imminent and poses serious risks if left unchecked.

Why Are Top AI Leaders Worried About Automation?

The researchers behind the "Pacing the Frontier" initiative argue that as AI systems become capable of automating their own development, the speed of progress could accelerate beyond society's ability to understand or control it. According to the statement, there is a significant risk that rapidly developed AI systems will become unintelligible and uncontrollable, making it difficult to predict how much faster automation will push the field forward.

"AI is advancing at a speed that our society cannot yet keep up with. At cutting-edge research institutions, we are on the verge of realizing AI that can surpass the most brilliant humans in almost every indicator of intelligence. This will bring unprecedented social and safety risks," stated Zhao Shengjia, chief scientist at Meta AI.

Zhao Shengjia, Chief Scientist at Meta AI

The signatories are not calling for a complete halt to AI development. Instead, they are requesting that the U.S. government support an international effort to develop the technologies and governance tools necessary to intentionally manage the pace of automated AI research.

What Specific Actions Are These Researchers Proposing?

The statement emphasizes the need for proactive measures before automation becomes widespread. John Shulman of Thinking Machines explained his reasoning for endorsing the initiative, noting that the goal is to foster a shared understanding of potential control mechanisms as AI automation research accelerates.

"I endorsed the statement to help foster a shared understanding of the potential need for control mechanisms as AI automation research accelerates dramatically. I hope that research institutes will begin designing such mechanisms independently, even before the U.S. government gets involved," said John Shulman.

John Shulman, Thinking Machines

How Can Governments and Institutions Prepare for AI Automation?

• International Governance Frameworks: Develop coordinated policies across countries to ensure that no single nation or company races ahead without safety guardrails, reducing the risk of a competitive dynamic that prioritizes speed over safety.
• Technical Control Mechanisms: Invest in research to create tools and systems that allow humans to monitor, understand, and intervene in automated AI development processes before they spiral beyond human comprehension.
• Independent Research Institution Initiatives: Encourage universities and research centers to begin designing control mechanisms now, rather than waiting for government action, to build a foundation of knowledge and best practices.

The timing of this statement is significant. The signatories believe that automation of AI research is not a distant possibility but an imminent development that requires immediate attention. By requesting government support now, they are attempting to establish frameworks before the technology becomes too advanced to manage.

Is There Skepticism About This Approach?

Not all voices in the AI community support the "slow down" narrative. The Tiny Corp, which develops AI hardware and inference engines, has publicly criticized the approach, arguing that some companies are calling for others to decelerate development while continuing their own work. The company pointed out the apparent contradiction in asking competitors to pause while maintaining one's own research momentum.

This criticism echoes broader concerns about the credibility of such statements. Elon Musk, for example, signed a statement in March 2023 calling for a six-month halt to AI development, yet was later revealed to be purchasing thousands of graphics processing units (GPUs) for his own AI projects. He subsequently founded xAI and released Grok, neither of which were part of the recent pacing initiative.

The contrast between public calls for restraint and private acceleration efforts highlights a fundamental challenge in coordinating AI development: without binding agreements and enforcement mechanisms, voluntary pledges may lack credibility. The researchers behind "Pacing the Frontier" appear to recognize this, which is why they are specifically requesting government involvement to create enforceable frameworks rather than relying on industry self-regulation alone.

The statement from over 1,100 AI professionals represents one of the most significant coordinated efforts by industry insiders to advocate for deliberate governance of AI development. Whether governments will respond with the kind of international coordination and technical investment the signatories are requesting remains to be seen, but the urgency of their message underscores growing concerns about the trajectory of artificial intelligence research.