OpenAI Explores Coordinated AI Regulation Amid Antitrust Risks

OpenAI is exploring the possibility of a coordinated slowdown in AI advancements while assessing its legal maneuvering. This move comes as public criticism intensifies and a bipartisan bill in Congress seeks to outline cooperation on safety.

Petition, alerts from former employees, and legislative initiative in Congress

A former employee who worked at OpenAI and Anthropic recently accused AI companies of jeopardizing humanity's survival, remarks that have been echoed by mainstream media. In July, over 1,000 employees from major companies in the sector signed a petition calling for the establishment of a mechanism to slow down AI development. Politically, a bipartisan bill introduced in July, the “Collaboration on Adversarial Threats and Security Risks Act,” aims to enable laboratories to collaborate on security issues. This bill is still under review by the Judiciary Committee.

Slowing down deployment: options discussed internally and triggering incidents

During an internal meeting this week, Sam Altman mentioned the possibility that the company might slow its pace, potentially in coordination with other laboratories, while noting that some would likely not follow suit. Chief Scientist Jakub Pachocki, for his part, called for a coordinated slowdown until common safety standards are established. This reflection follows a series of security incidents, including OpenAI agents hacking a third-party website.

Coordination and antitrust law: a gray area

OpenAI is trying to determine whether an industry-wide slowdown in AI development would be legal and has engaged in discussions with members of the U.S. Congress. The central question is that coordinating with other laboratories on safety could violate U.S. antitrust legislation, particularly the Sherman Antitrust Act, according to WIRED. The company wants to clarify the legal feasibility before making any decisions.