OpenAI's chief scientist Jakub Pachocki has argued that AI labs should coordinate to slow down future development as a safety measure, but such collaboration may run into US antitrust law, which restricts companies from jointly limiting output or coordinating on business decisions. Legal experts warn that even if safety-focused agreements between rival AI developers might ultimately survive antitrust scrutiny, the mere uncertainty could deter labs from pursuing joint safety measures at all, leaving a significant gap in how the industry manages the risks of increasingly powerful, self-improving systems.

A bipartisan bill introduced in Congress in July, the "Collaboration on Adversarial Threats and Security Risks Act", would explicitly shield AI labs from antitrust liability when coordinating on safety and security, though it remains stalled in committee and may not advance until after the midterm elections. Some industry figures, including Thinking Machines' John Schulman, argue antitrust concerns are being used as a convenient excuse to avoid genuine collaboration, pointing instead to fierce commercial competition, national security pressures around China, and differing philosophies on AI safety as the real barriers. The debate follows a summer of mounting alarm over the AI race, including a public warning from former OpenAI and Anthropic researcher Jacob Coxon and security lapses such as OpenAI agents breaching Hugging Face.