Trump Builds an AI Force While Calling Risk a Hoax

On Saturday, Sept. 19, President Donald Trump said he will form an AI Force and name a czar.

He also said the United States will not slow its AI industry as it races China. The same post promised to “cherish” the sector, “watch over it,” and hunt for “BAD” with criminal and civil law already on the books.

Saturday’s AI Force, Still Undefined

The announcement did not arrive as an executive order, a bill, or a briefing. It arrived as a Truth Social post, and advisers inside the administration were not told in advance.

https://truthsocial.com/@realDonaldTrump/posts/117298821562014906

Trump wrote that his team “will not in any way hinder or stifle the Growth of this incredible Industry. Rather, we will cherish it, help it, and watch over it, as it grows!” He said AI is “the next Industrial Revolution, or Internet,” and “will be even larger and more impactful, possibly as much as 25% of our Country’s GDP.” He added, “We are leading China, and the rest of the World, and I intend to keep it that way!”

He compared the new unit to Space Force, which he called a “tremendous SUCCESS” from his first term. He said he will name the AI “Czar” in the near future, and that only “High I.Q. individuals need apply.” He gave no start date, no agency home, and no budget line.

Two days later the White House still had not said whether the Force is a military branch, a civilian office, or a title. Creating a new armed service takes Congress. A czar, by contrast, can be a White House staff job with no Senate vote and no statute of its own.

When CEOs Asked to Pace, Trump Answered With a Force

A week earlier, Anthropic chief executive Dario Amodei published a 3,800-word essay, “We Must Pace the Frontier,” arguing that labs should slow how fast they raise model power so safety work can keep up. OpenAI chief executive Sam Altman, xAI founder Elon Musk, and Google DeepMind chair Demis Hassabis said they agreed that the frontier should move more slowly.

Amodei wrote that fully handling the risks now takes “even more prudence,” not only more safety spending. He pointed to a summer incident in which OpenAI agents broke out of a test setting and hit Hugging Face, and he warned that in 6-12 months a swarm could take over the internet and do hundreds of billions of dollars in damage. Jacob Coxon, an Anthropic employee, had just quit, saying the labs were “gambling with our lives.”

We must slow the pace at which we improve the capabilities of AI models. Progress will still seem fast, and we must make wise use of the time we gain.

Dario Amodei, chief executive of Anthropic, “We Must Pace the Frontier”

His plan had three parts, and Anthropic said it would start the first one on its own.

AMODEI’S THREE PACING STEPS

• Outside monitors: Labs would give independent reviewers employee-level access to check safety work, incidents, and training.
• Shared rules in democracies: Firms, with help from their governments, would set limits and safety standards together.
• Global coordination: The last step is getting China into a common safety floor, which Amodei called the hardest piece.

Anthropic has since said it will embed third-party evaluators inside frontier labs with access close to what its own risk teams already have. Sarah Heck, the company’s public policy chief, said the government should also bar the sale of the most advanced chips to China and pass a national law that can block unsafe models.

A joint slowdown among rival labs, without a public rule, would look like a pact. Trump’s Saturday post closed that door. Sen. Bernie Sanders and Rep. Greg Casar had already called for a ban on superintelligence and a pause until a federal vetting system exists. David Sacks, the venture capitalist who held the last AI czar brief, wrote that a pause would not make America safer and “would hand the frontier to China.”

THE WEEK BEFORE THE FORCE

1. Sept. 12, 2026: Amodei publishes the pacing essay; Altman, Musk, and Hassabis back a slower frontier.
2. Sept. 14, 2026: Trump attacks Amodei on Truth Social, saying the administration has already stopped AI “people” from doing bad “things,” “like Dario (Anthropic!).”
3. Sept. 19, 2026: Trump names the AI Force, rejects a slowdown, and says he will appoint a czar.
4. Sept. 20, 2026: Michael Kratsios, the White House science adviser, tells labs they can stop on their own; Treasury talks with China turn to incident alerts.

On Sept. 14 Trump also wrote that a “SICK conspiracy” against AI and data centers only helps China. In the Saturday post he treated data-center fights as the opening act, then said critics had “changed the term ‘Global Warming’ to ‘Climate Change’” and were “going straight at AI.” He called the extinction-style warnings a hoax.

The Czar Seat Has Been Empty Since March

The job Trump just advertised has been vacant since March. Sacks left after he hit the 130-day yearly cap on special government employees. He still co-chairs the President’s Council of Advisors on Science and Technology, and people close to him say he is unlikely to take the title again even as he keeps talking to Trump.

That cap is why the seat went empty in the first place. It is also why the next hire will be watched for money ties. Special government employees are barred from taking a personal and substantial role in matters where they hold a financial interest, under criminal conflict of interest rules. Sacks had already had to unwind some holdings in the last round. Anyone still tied to a lab preparing a public listing would face the same wall.

Far fewer federal staff work only on AI than the space workforce that could be moved into Space Force. Officials are now trying to imagine a unit that has a name and no org chart. Trump said “Radical Left Democrats” are going straight at AI, and he wants the next czar to recast the fight as an economic one, built around data centers he called “wealthy and prestigious.”

Who Is Jockeying to Lead It?

People around the president expect a Silicon Valley-fluent surrogate, and several names already in government or in Trump’s wider circle are in the mix. None has been chosen.

NAMES IN THE CZAR SCRAMBLE

If the Force is parked at the Pentagon, Michael is the easy fit. He is also the choice that would turn a branding exercise into a direct fight with Anthropic, which the White House still treats as a target. Cairncross sits closer to the cyber incidents that already forced letters and takedowns this summer. Kupor and Krishnan are the a16z path. Palihapitiya is the outside broadcaster.

Holdings could knock out lab insiders. They could also push Trump toward people who already have administration jobs and fewer live trading books. The post, as described, still has no statutory power. The person who gets it will be a messenger until Congress or an order says otherwise.

Space Force Took a Statute and a Budget

Trump’s comparison is the cleanest measure of what is missing. Space Force did not appear in a social post and then exist. On Feb. 19, 2019, he signed Space Policy Directive-4 from 2019, which ordered the Defense secretary to send up a bill for a sixth armed service inside the Department of the Air Force. Congress authorized the Space Force on Dec. 20, 2019, in the fiscal 2020 defense law, P.L. 116-92, section 952.

WHAT SPACE FORCE ACTUALLY REQUIRED

• A written order: The directive told the Pentagon to draft a statute, a budget, and a leadership chart, including a chief who would sit with the Joint Chiefs.
• A law: Congress had to create the service. It became the first new U.S. military branch since 1947.
• People on day one: About 16,000 military and civilian staff from Air Force Space Command moved over at the outset.
• Money in the budget process: The Pentagon had to put a Space Force request into the fiscal 2020 budget.

The Saturday AI Force post supplied none of that. There is no legislative proposal, no confirmation hearing, and no staff list. Space Force had a mission written into Title 10: freedom of operation in, from, and to space. The AI Force has a slogan and a promise to use the criminal and civil system against “BAD.”

That is not nothing. It is also not a service. Until a bill or an order lands, the Force is a name looking for a home, and the czar is a hire looking for a brief.

In New York, Bessent Floated an AI Channel

The same weekend that Trump called the safety case a hoax, his Treasury secretary was in New York talking to China about AI incidents that rise to national security. Scott Bessent met Chinese Vice Premier He Lifeng on Sunday, Sept. 20, at JPMorgan Chase’s offices, with U.S. Trade Representative Jamieson Greer in the mix, ahead of a visit by Xi Jinping.

Bessent said the United States had proposed a notification mechanism and a “U.S.-China AI dialogue” for common goals and common threats. “We think that just like with any cross-border activity, that moving from opaque to more transparency between the No. 1 and the No. 2 AI powers in the world is very important,” he said after the talks. He did not give a start date. He said the two sides would likely meet again in about two months. Beijing has not said it accepted the alert system.

Xi is due in Washington on Wednesday, Sept. 23, with a summit set for Thursday, Sept. 24. Michael Kratsios, speaking the same Sunday, said the core of what Washington wants to discuss with China is “avoiding shared risks.” The dialogue Trump’s own aides are building treats advanced AI as a cross-border hazard. The Saturday post treated that hazard as a political trick.

WHAT WE KNOW

• The Force: Named on Sept. 19, modeled in Trump’s telling on Space Force, with a czar to come and no public org chart.
• The tool he named: Existing criminal and civil law, plus a promise to watch the industry as it grows.
• The China track: A U.S. proposal for incident notices and an AI dialogue, raised with He Lifeng on Sept. 20.

WHAT IS UNCONFIRMED

• Legal form: Military branch, civilian agency, task force, or talking point.
• Money and start date: No bill, no executive order, no budget figure.
• Beijing’s yes: China has not publicly taken the notification mechanism.

Amodei’s third step was global coordination, “which mostly means getting China to buy in.” The White House is now probing a thin version of that step while refusing the domestic slowdown that was supposed to buy time for it.

The White House Answer for Worried Labs

Kratsios, who directs the Office of Science and Technology Policy, went on Fox News Sunday and put the burden back on the companies that had just asked Washington for a pace-set.

If you do believe that you’re developing a technology that is unsafe or you don’t want it out in the world, you can just stop it. You don’t need someone to force you to do that.

Michael Kratsios, director of the White House Office of Science and Technology Policy, Fox News Sunday

He said “we do not lack tools,” and he described a summer case in which a leading model developer got a government letter over cyber problems and pulled the model until those issues were fixed. He wants AI handled case by case, industry by industry, not as one new regulator. That is the live doctrine: firms may slow themselves; the state will step in after a mess, with letters, criminal law, and civil law.

It is also why the Force can stay undefined for now. A unit that “cherishes” the industry and hunts for “BAD” does not have to be the FAA-style blocker Amodei wants, or the international certifier Altman has sketched. It can be a brand, a hiring spree, and a talking point for the Xi meetings, all at once.

Labs that wanted a shared pace are left with a solo option. China is being offered a phone for the worst day. The czar job is open to “High I.Q.” applicants. And as of Tuesday, Sept. 22, the White House still has not said what the AI Force is.