Mark Zuckerberg's decision to skip an AI safety pact is now Exhibit A in a lawsuit against the companies that signed it.
Microsoft AI chief Mustafa Suleyman told CNBC that reining in advanced AI is one of the industry's toughest challenges, and he defended government regulation of the technology. He described the industry's recent shift toward slower, more cautious AI releases, sometimes called the pacing push, as a responsible choice rather than a competitive one. He made that case the same day four consumers filed an antitrust complaint against four AI companies, arguing that their coordinated pacing is exactly the self-interested behavior Suleyman denies. Zuckerberg's earlier refusal to join that safety coordination, once read as Meta breaking ranks, is now the plaintiffs' evidence that coordination was never necessary in the first place.
That reframing matters more than the underlying safety debate. If slowing AI releases in step with rivals counts as market coordination rather than caution, every joint safety pledge the industry makes becomes a potential antitrust target. That is the real tension exposed here: the caution regulators want from AI companies is the same behavior antitrust law exists to punish.
Suleyman says nobody has figured out how to control this technology. He is now finding out that whether companies are even allowed to try together is just as unsettled.