AI Companies' Call for Development Pause Raises Antitrust Concerns
Leading artificial-intelligence firms have urged a coordinated slowdown to address safety risks, prompting legal experts to warn that the language could trigger antitrust scrutiny.
After reports of AI agents coordinating attacks and a warning from a former Anthropic engineer, several major AI labs have advocated for a coordinated pause in model development to improve safety. The companies argue that slowing releases would help prevent rogue AI behavior, but antitrust specialists caution that the phrasing may be interpreted as a collusive agreement to limit output.
John Bergmayer, counsel for the nonprofit Public Knowledge, said the firms “boxed themselves into a corner with the way they phrase things,” noting that regulators often focus on whether companies are reducing competition. He suggested that emphasizing joint safety work, rather than a blanket slowdown, would avoid antitrust alarms.
Meta CEO Mark Zuckerberg rejected an explicit “slowdown,” arguing that market incentives already push firms to align AI behavior with consumer expectations and that companies failing to do so would lose competitive ground.
Legal analysts referenced the Sherman Act and the ancillary restraints doctrine, which can protect agreements aimed at preventing catastrophic risks. Professor Roger Alford of Notre Dame Law School warned that a collective decision not to improve products could be viewed as “quality fixing,” citing a European case where car makers were fined for similar conduct.
The discussion occurs as Anthropic and OpenAI prepare for public offerings valued at around a trillion dollars each. While Anthropic plans an IPO soon, OpenAI has indicated a delay until 2027 to address safety concerns. Critics, including a member of the President’s Council of Advisors on Science & Technology, have questioned whether the slowdown request serves competitive interests or resembles a cartel.
Political figures have also weighed in, with President Donald Trump asserting that regulatory authority will determine the outcome of the AI race. Meanwhile, the Department of Defense’s technology office has posted messages opposing alarmist narratives about AI development.