OpenAI Asks Congress Whether Slowing Down AI Development Could Be Illegal

OpenAI has recently approached members of the US Congress seeking guidance on whether an industry-wide effort to slow frontier AI development could run afoul of antitrust law, according to people close to the company who spoke to WIRED.

The question follows a blog post last weekend from OpenAI chief scientist Jakub Pachocki, who argued that coordinating a slowdown in future AI development is essential to keeping increasingly self-improving systems safe. He said he expects voluntary slowdowns to become common practice until the industry agrees on shared safety standards.

Legal Uncertainty

Some legal experts warn that such coordination could resemble companies agreeing to restrict output, potentially violating the Sherman Antitrust Act. Nicholas Felstead, a former AI policy fellow, has written that the outcome would depend entirely on the specifics of any agreement — but even if most safety collaborations would survive scrutiny, the mere uncertainty can discourage companies from trying.

In July, a bipartisan group of lawmakers introduced legislation that would explicitly shield AI labs from antitrust liability when coordinating on safety and security matters. The bill remains stalled in the House Judiciary Committee, and advocates say meaningful progress may not come until after the midterm elections.

More Than Just Legal Concerns

Critics argue that antitrust worries may partly serve as a convenient excuse, masking deeper reasons AI companies hesitate to collaborate: fierce competition for the emerging frontier-AI market, concerns about maintaining an edge over China, and sharp disagreements over the right approach to AI safety. John Schulman, an OpenAI cofounder now at rival lab Thinking Machines, publicly urged OpenAI and Anthropic to set aside their rivalry and jointly develop a pacing proposal, arguing that antitrust law does not actually prevent companies from drafting such plans together.

Concerns about the risks of unchecked AI competition have intensified this summer following several security lapses — including an incident in which OpenAI's own AI agents compromised Hugging Face — and a public warning from a former Anthropic and OpenAI researcher about the dangers posed by advanced AI systems. Lawmakers have increasingly called for stronger AI regulation in response.