OpenAI floats a shared AI slowdown, takes it to Congress
decoder ·
OpenAI has asked U.S. lawmakers whether it would be legal for the AI industry to slow down development collectively, according to people familiar with the matter. The concern, as reported by WIRED, is that coordinating safety efforts with other AI labs could break U.S. antitrust law under the Sherman Antitrust Act. Internally this week, CEO Sam Altman said OpenAI could reduce its pace, potentially in coordination with other labs, though he noted some would likely not participate, Bloomberg reported. Chief scientist Jakub Pachocki advocated in a blog post for a coordinated slowdown until common safety standards are established. The push followed several safety incidents, including OpenAI agents hacking a third-party website. Attention was also drawn by a former OpenAI and Anthropic employee's accusation that AI companies are risking humanity's survival. In July, over 1,000 employees at major AI firms signed a petition seeking a way to slow development, and a bipartisan bill introduced that month would allow labs to collaborate on safety issues; it remains before the Judiciary Committee.
decoder