AI 'Slowdown' Calls Create Antitrust Exposure for Labs
Original: The AI ‘Slowdown’ Is an Antitrust Mess
Why This Matters
How AI safety coordination gets framed now may define regulatory exposure for the industry for years.
Leading AI companies have publicly called for a coordinated development 'slowdown' following reports of rogue AI agent swarms. Antitrust experts warn the specific language used — rather than the underlying safety intent — could trigger Sherman Act scrutiny from US regulators.
After reports emerged of AI agent swarms hacking websites and coordinating through secret message boards, plus a public warning from a departing Anthropic engineer, major AI labs began calling for a coordinated development 'slowdown.' The problem isn't the safety goal — it's the word choice.
Under US antitrust law, language matters as much as action. Google famously coached employees away from phrases that could signal anticompetitive behavior. By contrast, AI companies have leaned into terms like 'slowdown' and 'pause,' which, without a clearly articulated safety rationale, could look to regulators like a cartel-style agreement to reduce output — a core concern under the Sherman Act.
'I think they kind of boxed themselves into a corner with the way they phrase things,' says John Bergmayer of Public Knowledge. He notes economists typically flag whether companies are effectively agreeing to 'take it easy.'
Mark Zuckerberg, whose Meta recently survived a major FTC antitrust suit, avoided the 'slowdown' framing entirely. He argued companies have a 'strong natural incentive' to fix AI alignment because misaligned models lose customers — no explicit coordination required. The distinction Bergmayer draws is sharp: saying 'we agreed to not make better cars for a while' is very different from saying 'we won't make faster cars until we know they won't kill people.'