OpenAI Asks Congress if Coordinated AI Slowdowns Are Legal
OpenAI has asked Congress whether coordinating an industry-wide development slowdown would violate antitrust laws, highlighting a major legal hurdle for collaborative AI safety efforts.
OpenAI recently lobbied congressional lawmakers for guidance on whether orchestrating a collective pause on frontier AI development could trigger antitrust violations. The inquiry follows a blog post by OpenAI chief scientist Jakub Pachocki, who advocated for "coordinating to slow down future development" to ensure the safety of self-improving systems. Pachocki noted that voluntary slowdowns might become necessary until shared safety benchmarks are established, but legal experts warn such agreements could run afoul of the Sherman Antitrust Act by restricting market output.
Nicholas Felstead, an assistant director at the Australian Competition and Consumer Commission, previously warned that while safety collaborations might survive scrutiny, legal uncertainty acts as a powerful deterrent. In response to these concerns, a bipartisan group of U.S. lawmakers introduced the Collaboration on Adversarial Threats and Security Risks Act in July. This bill would explicitly shield AI laboratories from antitrust liability when cooperating on safety and security. Caleb Knapp, director of government affairs at the AI Policy Network, indicated that congressional appetite to pass the bill is growing, though action may wait until after the midterm elections.
Some industry figures remain skeptical of the antitrust defense. John Schulman, an OpenAI cofounder who now serves as chief scientist at rival lab Thinking Machines, argued on social media that antitrust concerns are a "fake" excuse. Schulman asserted that antitrust laws do not prohibit companies from jointly developing safety proposals. Critics suggest that fierce commercial competition, national security concerns regarding China, and differing safety philosophies are the true barriers to collaboration.
For AI practitioners and safety researchers, this legal debate directly impacts how organizations share threat intelligence and coordinate safeguards. Recent incidents, such as OpenAI agents hacking Hugging Face, have intensified calls for industry-wide standards. Until Congress provides clear safe harbors, developers must navigate a complex landscape where collaborating on safety protocols could be misconstrued as anticompetitive collusion, potentially delaying the deployment of unified defense mechanisms.
This is our own summary of reporting by WIRED AI
