OpenAI has reached out to members of the U.S. Congress to clarify whether a coordinated slowdown in the development of advanced AI models would violate antitrust laws. This was reported by WIRED, citing sources close to the company.
The issue arises because OpenAI, Anthropic, Google, and other industry players are competitors. An agreement among them to slow development could theoretically be seen as a joint restriction of product output.
Nicholas Felstead, Deputy Director of the Australian Competition and Consumer Commission and former researcher at the Center for Law & AI Risk, noted that such coordination could potentially conflict with the Sherman Act. However, the legal assessment depends on the specific terms of the agreement.
“Even if most forms of safety cooperation ultimately withstand antitrust scrutiny, legal uncertainty itself can be a powerful deterrent,” Felstead wrote.
Not all industry representatives find this argument sufficient. OpenAI co-founder and Chief Scientist of Thinking Machines, John Schulman, urged OpenAI and Anthropic to jointly develop a proposal for AI development pace. In his view, antitrust laws prohibit certain agreements between competitors but do not prevent them from jointly preparing such a mechanism.
Exemption for AI Companies
On July 23, lawmakers from both parties introduced the Collaboration on Adversarial Threats and Security Risks Act. The initiative was proposed by Senators Adam Schiff and Jim Banks, along with Representatives Bob Latta and George Whitesides.
The bill provides a limited antitrust exemption for collaborative work on AI risks. Developers will be able to share threat information and coordinate measures to mitigate them, including delaying or restricting the deployment of high-risk models.
In cases of coordinated restriction, companies must notify the head of the U.S. Department of Justice’s antitrust division in advance, specifying the particular risk and scope of the proposed measures. The exemption does not permit price fixing, market division, boycotts, or other anti-competitive practices.
Both versions of the initiative remain in the legal committees of Congress. H.R. 9914 was sent to the House Judiciary Committee, and S. 5105 to the Senate. No further actions have been recorded since their introduction on July 23.
OpenAI Has Already Slowed Development
On September 6, OpenAI’s Chief Scientist Jakub Pachocki wrote that no lab has yet solved alignment and monitoring issues well enough to continue scaling at maximum speed for an extended period.
I wrote about the state of AI, why I’m concerned about the next few years, and the choices we need to make to keep the future in humanity’s hands.
An Alien Mind: https://t.co/FeIfWNe0UE
— Jakub Pachocki (@merettm) September 6, 2026
“I expect and hope that voluntary slowdowns will become common practice until general safety standards emerge,” he noted.
In his opinion, international coordination of further AI development should become a priority for governments.
Previously, OpenAI temporarily slowed the scaling of new systems and paused reinforcement learning for the latest models intended for deployment for two weeks. The decision followed an incident with Hugging Face and a preliminary assessment of Astra’s cyber capabilities.
Later, researchers noted a decrease in Astra’s observability. Pachocki then acknowledged this monitoring capability as “fragile.”
Earlier, in September, a U.S. Senate subcommittee began an investigation into OpenAI’s actions following the Hugging Face incident. Lawmakers demanded documents on the company’s response to the unforeseen behavior of its AI agents.
Before this, there was a proposal in the U.S. to permanently ban the development of artificial superintelligence and temporarily halt the creation of advanced AI systems until federal safety standards are established.
