OpenAI has approached members of Congress for guidance on the legality of a potential industry-wide slowdown in frontier AI development, according to sources familiar with the matter. Coordination on safety among AI labs may conflict with antitrust laws, which could hinder collaboration among major tech companies.
Jakub Pachocki, OpenAI’s chief scientist, recently published a blog post advocating for a coordinated slowdown in AI development to ensure the safety of self-improving AI systems. He anticipates that voluntary slowdowns may become common until safety standards are established.
Legal experts, including Nicholas Felstead from the Australian Competition and Consumer Commission, have expressed concerns that such coordination could violate the Sherman Antitrust Act, depending on the specifics of any agreement. Felstead noted that while most safety collaborations might withstand antitrust scrutiny, legal uncertainty could deter companies from participating.
OpenAI did not provide comments prior to publication.
There are indications that Congress is responding to these concerns. In July, a bipartisan group of lawmakers introduced the “Collaboration on Adversarial Threats and Security Risks Act,” aimed at allowing AI labs to collaborate on safety without risking antitrust violations. The House version of the bill has been referred to the Judiciary Committee but has not yet been addressed.
Caleb Knapp, director of government affairs at the AI Policy Network, which supports the bill, stated that it would facilitate legal cooperation among AI labs to tackle safety issues. He mentioned that Congress is increasingly interested in addressing AI safety, but legislative action may be delayed until after the midterm elections.
While some AI executives express concerns about antitrust implications, others believe these worries serve as a cover for deeper issues that hinder collaboration among AI developers. The competitive nature of the AI market and differing opinions on safe AI development contribute to this reluctance.
John Schulman, an OpenAI cofounder and chief scientist at Thinking Machines, suggested that industry leaders should collaborate on a pacing proposal despite citing antitrust concerns.
Concerns about the competitive race among AI companies have gained national attention, particularly following warnings from former researchers about the risks posed by rapid AI advancements. Recent security incidents, including OpenAI’s agents hacking Hugging Face, have raised alarms about the adequacy of industry safeguards, prompting calls for AI regulation from lawmakers.