On Saturday, President Trump announced on Truth Social that he is forming an "AI Force," modeled on the Space Force he created in his first term, and will soon name a new AI czar. Only "High I.Q. individuals need apply," he wrote. He offered no budget, no structure, no mandate, and no explanation of where this Force would sit in the federal government.
What he did offer was clarity of a different kind, and it is the part worth reading twice. "We will not in any way hinder or stifle the Growth of this incredible Industry," he posted. "Rather, we will cherish it, help it, and watch over it, as it grows!" Bad behavior, in his telling, will be handled by "our already existing Criminal and Civil Justice System."
Strip away the branding and that sentence is the entire policy. There will be no federal rules written before AI systems cause harm. There will be prosecutions and lawsuits after they do. If that governance model sounds familiar, it should. It is roughly how the United States handled social media for two decades, and the country is still litigating how that worked out.
The timing makes the message impossible to miss. Exactly one week earlier, on September 12, Anthropic CEO Dario Amodei published an essay titled "We Must Pace the Frontier," urging industry-wide cooperation to slow development in favor of stronger safety measures, and committing his company to giving independent evaluators permanent, employee-level access inside its labs. OpenAI's Sam Altman, SpaceXAI's Elon Musk, and Google DeepMind's Demis Hassabis publicly agreed the same day. Altman told Fortune the leading labs were close to a formal pact to slow down collectively.
These are not outside critics. These are the people building the systems, and they have been sounding increasingly alarmed. An Anthropic researcher, Jacob Coxon, recently resigned in public, accusing both Anthropic and OpenAI of "racing straight to self-improving superintelligence and gambling with our lives." Anthropic's own alignment science lead, Evan Hubinger, responded that many researchers inside the two companies genuinely believe advanced AI could kill everyone, and Axios reports Hubinger personally puts the odds of that outcome within a decade at greater than 10 percent. The warnings followed disclosures of autonomous hacking incidents during testing, serious enough that California's governor cited a loss-of-control attack on the platform Hugging Face in a new executive order.
Trump's response to all of this has been to call the fears a "hoax" and to say the only guardrail AI needs is a strong and smart president. The AI Force is that position with a logo attached.
Now consider what the industry actually asked Washington for, because it makes the answer even stranger. Amodei, aware that rival CEOs coordinating a slowdown could violate antitrust law, asked the government to mediate or at least issue "a narrow waiver for certain kinds of safety conversations." Altman said OpenAI would welcome a federal framework with consistent safety requirements. In other words, the labs requested a referee and a rulebook. The White House declined to provide either, and then the civil justice system Trump pointed to showed up right on schedule, aimed in the opposite direction. On Friday, paid subscribers sued Anthropic, OpenAI, Google, and SpaceXAI in federal court in San Francisco, arguing their agreement to slow development is an illegal price-fixing-style conspiracy that devalues ChatGPT, Claude, Gemini, and Grok subscriptions. The plaintiffs' lead attorney, Nick Rowley, framed the suit as pro-safety, arguing that existential risk decisions shouldn't be controlled by private agreements among for-profit companies. Whatever the merits, the practical effect is hard to ignore: the first major legal action of the AI slowdown era is an attempt to make the slowdown illegal.
That leaves the labs in a vise. A July statement signed by high-ranking employees across the major labs acknowledged the "intense competitive pressure not to unilaterally slow" development, which is precisely why the CEOs sought coordination and legal cover. Coordination is now under antitrust attack. A federal framework has been refused. The fallback, per the president, is that every company slows down alone or not at all, and everyone finds out later what the damages were worth.
There is also a contradiction worth naming inside the administration itself. While the White House promises to cherish the industry, the Commerce Department temporarily imposed export restrictions on Anthropic's most advanced models this year, and the Defense Department tried to blacklist Anthropic as a supply-chain risk until a judge ruled the move illegal. One arm of the government is hugging AI while two others swing at its most safety-conscious company. You can call that a balanced approach if you are feeling generous. A less generous reading is that there is no AI policy, only competing impulses, and the AI Force exists to paper over the gap.
Even the Space Force comparison undercuts the announcement. Space Force is a real military branch because Congress passed a law creating it, funding it, and defining its mission. No bill, budget line, or agency home has been announced for the AI Force, and Politico reports the industry was not briefed before the post went up. David Sacks, the previous AI and crypto czar, stepped down earlier this year to co-chair the president's Council of Advisors on Science and Technology, and Axios notes the White House had not planned to refill the czar role six months ago. Until a person, a budget, and legal authority exist, the AI Force is a promise of acceleration wearing the costume of oversight.
The vacuum is already being filled from below. The day before Trump's announcement, California Governor Gavin Newsom signed an executive order accelerating two state AI safety laws endorsed by Anthropic and OpenAI, and directing a task force to develop recommendations on independent audits and a potential kill-switch requirement for frontier models. "The federal government's abject failure to create any form of meaningful AI oversight or accountability should alarm every American," Newsom said, "especially when AI CEOs themselves are begging for regulation."
Meanwhile the politics on the ground cut against the cheerleading. Gallup found this spring that 71 percent of Americans oppose having an AI data center built near them, with nearly half strongly opposed, largely over water, energy use, and utility costs. Data centers have become a live midterm issue, which Trump acknowledged in the same post by claiming the backlash has mostly failed. The public is voting on this technology's footprint whether Washington writes rules or not.
What happens next is fairly predictable. The antitrust lawsuit will test whether safety coordination survives without the federal waiver the industry requested. States will keep building a patchwork of rules the White House dislikes. And the AI Force will either acquire a budget, a leader, and a mission, or join the long list of announced entities that amounted to a name.
Free Game Takeaway
Watch three concrete signals. First, the antitrust suit in the Northern District of California: if courts treat safety coordination among AI labs as illegal collusion, and Congress never grants the narrow waiver Amodei requested, the slowdown pact collapses and the race dynamic the labs' own employees warned about takes over. That directly affects the pace of the tools reshaping your job and your feed. Second, your state capital, not Washington, is where actual AI rules will be written in the near term; California is already moving on audits and kill-switch requirements, so the protections you get may depend on your zip code. Third, if you live anywhere near a proposed data center, the Gallup numbers (71 percent local opposition, driven by water and electricity costs) tell you this is now a potent local political issue with midterm leverage, and showing up at zoning and utility hearings is currently the most effective form of AI regulation available to an ordinary person. For investors watching the AI buildout, the absence of a federal framework does not mean low policy risk; it means the risk has shifted to courts and 50 statehouses, which is harder to price and slower to resolve.