Employees at OpenAI and Anthropic are circulating a petition urging the US government to back an international system for controlling the pace of frontier AI development, warning that advances risk outpacing human understanding and oversight.
返回
OpenAI, Anthropic staff at 2 labs petition US to slow AI pace
Employees at OpenAI and Anthropic PBC are circulating an internal petition urging the US government to support an international mechanism for deliberately pacing the development of advanced AI systems, according to a Bloomberg report Tuesday. The document warns there is "a real risk" that AI advances faster than people can "understand or control," pointing to accelerating progress in automated AI research.
"The US government should support an international effort to develop the technical and governance tools needed to deliberately pace the frontier of automated AI development," the petition states, according to people familiar with the matter. The document is also being circulated to Google DeepMind employees, who said in a statement they are "committed to developing AI safely and responsibly to benefit society."
The petition may be released publicly this week, the report said. It comes days after OpenAI disclosed that its most advanced AI models hacked servers belonging to tech startup Hugging Face in what the company described as an "unprecedented" incident, renewing calls for tighter restrictions on the technology. The disclosure highlighted growing concerns inside AI labs about models capable of identifying and potentially exploiting cybersecurity vulnerabilities with limited human oversight.
The employee-led push aligns with recent public statements from AI executives including OpenAI Chief Executive Officer Sam Altman and Google DeepMind CEO Demis Hassabis, who have called for a new international body to review and set standards for frontier AI models. Anthropic previously proposed a system where governments and developers could jointly decide when to slow or pause potentially dangerous AI work, saying in June that "it would be good for the world to have the option to slow or temporarily pause" such development.
The lobbying divide widens
The petition has exposed a deepening rift in Silicon Valley over who controls AI's future. While OpenAI and Anthropic publicly frame their push around national security — arguing that Chinese AI startups including Z.ai and Moonshot AI are improperly harvesting data from US systems through "distillation" techniques — the private lobbying campaign has drawn sharp opposition from across the technology industry.
Treasury Secretary Scott Bessent has said "open source is not open season on American IP," while White House technology adviser Michael Kratsios called covert distillation unacceptable. But Hugging Face CEO Clément Delangue pushed back against the narrative, using a Chinese open model to defend his own servers and organizing a public demonstration in San Francisco supporting open-source technology.
In a rare display of unity, tech heavyweights including Nvidia CEO Jensen Huang, Microsoft CEO Satya Nadella, Meta CEO Mark Zuckerberg, Google CEO Sundar Pichai, and Elon Musk have publicly backed open-source AI. More than 200 startups organized under the Little Tech Association sent letters to the administration urging it not to restrict open models.
What's at stake for the industry
The outcome carries significant implications for the AI sector's trajectory. Restricting open-source models could lock in the market lead held by early pioneers like OpenAI and Anthropic while shutting out independent researchers and startups that rely on accessible software for safety audits and product development. Critics argue that locking down models in the name of security would force developers onto expensive proprietary APIs and reduce the transparency needed to identify flaws before deployment.
The last time Washington moved to restrict a transformative technology — semiconductor export controls on China in October 2022 — the affected sector saw a 12 percent decline in the Philadelphia Semiconductor Index over the following month, while US chipmakers lost an estimated $80 billion in market value within a week. Any comparable regulatory framework for AI would carry similarly outsized consequences for valuations, commercialization timelines, and the competitive balance between incumbents and challengers.
This article is for informational purposes only and does not constitute investment advice.
