AI Pause Plans Face Warnings of Monopoly and Foreign Risk

WASHINGTON — Sen. Bernie Sanders, I-Vt., and Rep. Greg Casar, D-Texas, are leading legislative efforts to halt development of advanced AI systems until federal safety regulations are established. Their proposal also calls for a permanent prohibition on creating artificial superintelligence, described as hypothetical AI systems whose cognitive capabilities surpass human intelligence.

The effort comes amid broader federal attempts to establish AI guardrails. President Joe Biden’s October 2023 executive order requires developers of large foundation models to submit safety test results to the federal government when their training process exceeds specific computational thresholds.

Jennifer Huddleston, a technology policy scholar at the Washington-based Cato Institute, wrote in a policy post released Monday that mandatory pauses would do little to enhance safety while severely harming market dynamics. “A government-mandated pause would raise several concerns, and likely fail to fulfill its alleged safety improvements,” Huddleston wrote.

She said federal rulemaking moves slowly compared with technological change, warning that “a regulatory framework built for today’s models may actively hinder tomorrow’s, preventing what could be better or safer responses.” Huddleston also contended that market leaders could use safety negotiations to consolidate power and create high barriers to entry for smaller startups.

Jack Dorsey, chairman and co-founder of financial technology firm Block and co-founder of Twitter, raised a similar concern Monday about proposals negotiated directly between government officials and established AI leaders. “The companies leading machine intelligence deserve to be heard. They have expertise and commercial interests to protect,” Dorsey posted on X. “Rules built around their resources could make them the only ones able to participate. A sincere concern about safety can still produce a barrier to entry.”

Free-market advocates and open-source champions have warned that federal intervention could produce regulatory capture, in which rules crafted in consultation with top industry players inadvertently insulate those firms from competition. Dorsey said safety should instead be pursued through open releases that allow independent researchers to inspect, modify, and test models.

He urged developers to publish evaluations and known limitations so external experts can verify safety claims. Although he acknowledged that high computational capacity could trigger independent scrutiny, Dorsey opposed blanket approval requirements, general waiting periods, or limits negotiated behind closed doors by dominant tech firms.

“Preserving a company’s commercial advantage is not a safety objective,” Dorsey wrote. He added that mandatory publication delays should be limited to cases where evidence shows an open release would directly increase catastrophic risks that narrower remedies cannot contain.

Huddleston separately cautioned that a unilateral pause on U.S. technology development could undermine national cybersecurity defenses while foreign rivals, including China, rapidly advance their own AI capacities. Policy analysts at the Atlantic Council have also emphasized that competitive market pressures make enforceable safety standards necessary, while deep geopolitical distrust between the U.S. and China impedes international treaties or joint enforcement agreements.

Major AI developers, including OpenAI, have publicly supported slower development timelines and encouraged companies to adopt voluntary safety protocols. The White House secured voluntary commitments from major AI developers in July 2023 to undergo third-party testing and watermarking of AI-generated content.

OpenAI CEO Sam Altman has backed slower development and urged companies to strengthen safeguards without waiting for federal rules. OpenAI has also asked lawmakers whether competing AI firms could legally agree among themselves to pause or slow development without violating federal antitrust statutes, including the Sherman Act, which prohibits agreements that restrict trade or limit output.

At the state level, California’s proposed SB 1047 created similar tensions in 2024. The bill would have required safety testing and emergency shutdown mechanisms for large-scale AI models, but California Gov. Gavin Newsom vetoed it after opposition from open-source developers, academics, and venture capitalists who argued that it would suppress technological innovation and drive companies out of the state.