OpenAI has requested clarification from the US Congress regarding whether coordinating the slowing of AI development among competitors could violate antitrust laws. The company expressed concerns that such agreements might be interpreted as an unlawful restriction of product output.
Legal experts note that such coordination could potentially conflict with the Sherman Act. While some industry figures, such as OpenAI co-founder John Schulman, argue that companies could jointly propose development mechanisms without violating competition rules, legal uncertainty remains a significant barrier.
On July 23, lawmakers introduced the Collaboration on Adversarial Threats and Security Risks Act. The bill proposes a limited antitrust exemption allowing developers to share threat data and coordinate the delay or restriction of high-risk AI model deployment, provided they notify the Department of Justice.
The proposed legislation, currently in committee, prohibits price-fixing, market division, and boycotts. Meanwhile, OpenAI has already implemented voluntary slowdowns, including a two-week pause in reinforcement learning for new models following security concerns related to its Astra system.
The US Senate subcommittee is currently investigating OpenAI's internal processes following an incident involving Hugging Face. OpenAI's chief scientist Jakub Pachocki has publicly advocated for voluntary development slowdowns as a standard practice until safety benchmarks are established.
