Four plaintiffs filed a federal class action Friday in the U.S. District Court for the Northern District of California accusing Anthropic, OpenAI, SpaceXAI, and Google of colluding to slow frontier A.I. development, an antitrust theory that reframes the industry’s safety turn as an illegal restraint of trade. One day later, President Trump used Truth Social to announce an “AI Force, much like I did Space Force,” alongside a new A.I. czar. The two events, arriving inside twenty-four hours, sketch the political geometry of the next year: a courtroom pushing the labs toward faster release, a White House pushing in the same direction, and a Senate that isn’t inclined to grant anyone permission to slow down.

The complaint traces its evidence to a September 12 essay by Anthropic chief executive Dario Amodei calling for “industry-wide coordination” to “pace the frontier.” Within hours, Elon Musk of SpaceXAI posted “Dario is right,” Sam Altman of OpenAI publicly agreed, and Demis Hassabis of Google DeepMind called it “the right path forward.” Amodei’s essay had also asked Washington for “a narrow waiver for certain kinds of safety conversations,” an unusual acknowledgment that the conversation he was proposing might otherwise be illegal.

The plaintiffs’ theory is direct. “The antitrust laws do not permit competitors to decide among themselves that competition is too dangerous,” the filing argues. Lead attorney Nick Rowley told CBS News that A.I. “could kill us all” if safety protocols are left to “private self-serving agreements,” a lede designed for the class-certification cycle as much as the merits.

At a Senate hearing Tuesday, Josh Hawley rejected Amodei’s waiver idea in five words: “Absolutely no way that’s happening.”

Trump’s Saturday post supplied the political echo. “We will not in any way hinder or stifle the Growth of this incredible Industry,” he wrote, adding that bad actors would be handled by “our already existing Criminal and Civil Justice System.” The White House hasn’t clarified whether the AI Force would be a military branch, per CNN, and the administration is expected to convene a high-level A.I. event Wednesday on the sidelines of the U.N. General Assembly.

The subtext is familiar to anyone who watched the recently scheduled Washington–Beijing A.I. safety talks get announced this month: safety is now a diplomatic vocabulary the labs can’t control. ChatGPT, Claude, Grok, and Gemini remain on shelf; the question is whether the pace of what they can do is set by the four CEOs, by a judge in San Francisco, or by an executive branch that reads any hesitation as unpatriotic. For now, all three are pulling in different directions, and none of them is pulling toward pause.

Sources