The delegation, which included senior figures from OpenAI, Anthropic, Google and Meta, urged the United States government to support an international framework designed to “deliberately pace” the development of advanced AI systems. The proposal, outlined in a briefing document shared with congressional staff, calls for coordinated oversight to mitigate what the signatories described as catastrophic risks posed by increasingly capable models.
The appeal marks a significant shift in an industry that has long championed rapid iteration and minimal regulation. “We are moving faster than our ability to understand the consequences,” one of the researchers involved in the briefing told The Chronicle Page on condition of anonymity, citing the sensitivity of ongoing discussions.
The proposed framework would not halt development entirely but would introduce mechanisms for monitoring, testing and potentially pausing the release of systems that exceed certain capability thresholds. The group specifically cited risks including the potential for AI systems to be used in the development of biological weapons, cyberattacks at scale, and the loss of human control over autonomous decision making.
Washington has so far taken a fragmented approach to AI governance. The Biden administration issued an executive order in October 2023 requiring safety testing from leading developers, but legislative efforts in Congress have stalled amid partisan disagreements over the scope of regulation. The new push from industry insiders could provide fresh momentum for a more comprehensive federal framework.
Not all major AI companies have signed on to the effort. Some smaller developers and open source advocates have argued that calls for international pacing are a veiled attempt by market leaders to entrench their dominance by raising barriers to entry. The briefing document acknowledged these concerns, stating that any framework must “preserve the benefits of open research while addressing systemic risk.”
The group plans to present a more detailed proposal at the upcoming International AI Safety Summit in London later this year. European regulators, who have already passed the bloc’s sweeping AI Act, have expressed interest in coordinating with the United States on cross border standards. Whether the current administration or a future one will take up the industry’s invitation remains an open question.
