On Sunday, October 4, 2026, President Trump announced what he calls the Superintelligence Force, a White House task force to coordinate federal AI policy. Two days later, the hosts of the podcast Moonshots with Peter Diamandis and their guest, Emad Mostaque, tried to work out what it means. The panelists welcomed the government's attention to advanced AI. They differed over whether a body like this can steer it.
What the task force is
Peter Diamandis, the founder of XPRIZE and Singularity University, opened with a summary of the announcement. The task force is chaired by Jay Clayton, the director of national intelligence, whom Diamandis said the White House had named its "AI czar" just two months after he became "America's top spy." Diamandis listed Federal Trade Commission chair Andrew Ferguson, Pentagon chief technology officer Emil Michael and the director of the Office of Personnel Management among the members. He said the group reports to the president and to chief of staff Susie Wiles.
The membership is wider than that. Semafor reported that it also includes the vice president, the defense secretary, the Treasury secretary and David Sacks. Semafor also said the task force has 120 days to produce a report on AI's risks, opportunities and federal responsibilities, matching Diamandis's account. Diamandis said the report is to cover AI incident reporting and whether the government could respond to a serious incident. He read from the president's post, which said the force will coordinate the government's dealings with "consumers, public interest groups, religions, critical infrastructure providers, and superintelligence companies."
Ismail: a 20th-century tool for a 21st-century technology
Salim Ismail, founder of Open ExO, said he was glad the government is taking superintelligence seriously. "But the name gives away the problem," he said. In his view, the task force assumes that AI can be centrally coordinated.
"So you're creating a 20th century classic task force to address a distributed 21st century technology," Ismail said. He called it "a fundamental impedance mismatch," borrowing an electrical-engineering term for two connected parts that are badly matched, so that much of what passes between them is lost.
Ismail said the government should stop debating whether the technology should advance. Instead, it should build "scaffolding" around AI: incident reporting, identity and liability mechanisms that help guide where it goes. He said the bigger problem is that officials treat the risk of not being first as high, which he said is what Clayton said. His alternative is "bounded experimentation": move quickly, "instrument everything," publish failures and settle accountability fast. "You can't centralize or manage centrally an intelligence explosion," he said. "What you can do is make sure the blast radius is a bit more bounded."
Wissner-Gross: from 1939 to 1942
Alexander Wissner-Gross, a computer scientist and founder of Reified, read the move as acceleration. He said the president has repeatedly argued that whoever wins at AI wins, and that putting Clayton in charge fits that view.
About a year ago on the show, Wissner-Gross said, he compared that moment to 1939 in the history of the Manhattan Project, the American program that built the atomic bomb. "I think we're speedrunning the Manhattan Project for superintelligence," he said. "I think this is maybe now a 1942 moment."
The date has a specific meaning. A history from the National Nuclear Security Administration describes 1942 as the year the project's organization changed: the Army took responsibility for the plants that would produce bomb material, the weapons laboratory and security. General Leslie Groves received overall command on September 17, 1942, and quickly secured uranium supplies and major sites: Oak Ridge for uranium processing, Los Alamos for weapons design and testing, and Hanford for plutonium production.
Wissner-Gross said that putting the nation's top intelligence official in charge of a single task force, "no pun intended," designed to remove obstacles to AI acceleration sends a clear signal. The audience, he said, includes members of Congress "who might have decelerationist tendencies," industry and, most importantly, foreign governments. The message is that the U.S. "intends to accelerate, not decelerate." He added that even without the warfare connotations, "this is a World War II moment. And the U.S. intends to win it with its own Manhattan Project."
Blundin: what kind of race, and who decides
Dave Blundin, founder of Link Ventures, said the effort is being characterized, and treated, as a race. He then pointed to Alvin Graylin, a friend of the show, who Blundin said asks how this can be a race when it "goes to infinity for the next millennium." If you cross some threshold and declare victory, Blundin asked, does everything stop that day? He said many race scenarios involve suppressing other civilizations, "which doesn't sound like an ideal thing." Against that background, he called bringing religious leaders into the conversation "a pretty good idea."
His other concern was how the policy is made. Blundin said he had spoken to the chief executive of a public company, whom he did not name, just before recording. With midterm elections coming, he put the chance that Congress flips to the other party at 75%. Yet these actions come as executive orders from the White House, he said, "with nothing to do with Congress whatsoever," as do international negotiations and tariffs. He called that "very unusual in history."
The view from Europe
When Diamandis asked about the mood across the Atlantic, the discussion described Europe as very sleepy and still focused on execution-based AI, with efforts to reach the frontier largely abandoned. As evidence, it cited Mistral, the French AI company, having just released a top model scoring 38 on Artificial Analysis, against about 66 for the top American models. Artificial Analysis is an independent firm that ranks models. Its Intelligence Index combines ten evaluations covering agent tasks, coding, scientific reasoning and general tasks, and mainly measures English text. Europe, the discussion suggested, is mostly about execution and regulation, with little drive toward superintelligence.
The likely exception was defense, where interest was described as picking up dramatically, as in the U.S. Diamandis remarked that this year's global AI spending roughly equals the world's total defense budget. The conversation went on to say that AI spending will keep rising and has not yet touched the defense budget, noting that an aircraft carrier or an F-35 fighter costs about as much as training frontier models.
The discussion also turned to the other side of AI that governments have noticed: surveillance on a scale that was impossible because you couldn't keep eyes on everyone until now. It forecast near-perfect AI lie detectors soon, and AI inside machines everywhere that would make the Five Eyes intelligence alliance look like nothing. Even so, it was suggested that Clayton's role as director of intelligence makes sense, because America would not go down the dark path if it can avoid it. The expectation was of some push for liberty by the government, though future governments were an open question. In other countries, spying on their own people was expected to go crazy. The UK was named as a place to watch: given its monitoring of social media, it was described as susceptible, with a dark path there for individual liberty.