Zendoric
← Back to the day · October 5, 2026

Trump's AI czar is a spy chief: the Superintelligence Task Force files AI under security, not consumer law

🕒 Published on Zendoric: October 5, 2026 · 00:06

✨ AI-generated · how it's made

Trump has put Jay Clayton, his director of national intelligence, in charge of a White House AI task force with 120 days to report. The roster — intelligence, defense, technology, personnel — tells you more than the name does: Washington has decided AI is a national-security file, not a consumer-protection one. That choice buys speed and costs durability.

President Donald Trump has named Jay Clayton, his director of national intelligence, to help lead a new White House task force on artificial intelligence, according to Bloomberg. Trump announced it in a social-media post, and gave the group a name that is itself a policy statement: the "Superintelligence Task Force." It has 120 days to produce a report, the Wall Street Journal reported.

Clayton is not alone at the top. Trump said the other leaders are Andrew Ferguson, described in the report as a deputy secretary at the Pentagon; chief technology officer Emil Michael; and Scott Kupor, director of the Office of Personnel Management — the agency that manages the federal workforce. All of them report to Trump and to White House chief of staff Susie Wiles. Clayton has overseen the sprawling US intelligence bureaucracy since his Senate confirmation in July, and the Journal had already reported he was expected to be named AI czar; chairing the task force performs that function in practice. Bloomberg adds a detail worth keeping in view: he is a friend and golf partner of the president's.

The mandate, in Clayton's words as quoted by the Journal, is to ensure "that we remain leaders in superintelligence and that the interests of the American people come first." Trump has rejected calls for new regulation and has instead backed a voluntary framework built on external safety audits and tighter internal controls. "We have a great lead, and we're going to keep it," he said last week at a meeting that, per Bloomberg, brought nearly two dozen Silicon Valley executives to the White House, among them Nvidia's Jensen Huang and Anthropic's Dario Amodei. Treasury Secretary Scott Bessent pushed harder in an Axios interview published 3 October, calling prominent AI leaders' existential-risk warnings alarmist and unhelpful and placing responsibility on "the people who run the labs" — adding that he believes the labs have reached the same conclusion.

Our thesis: the roster is the policy. Four names, and not one of them is a regulator in the ordinary sense. Intelligence, defense, technology, federal personnel — the desks chosen are the ones that handle secrets, weapons, procurement and hiring. That framing determines which levers get pulled. Not statute, but export controls, classification, security clearances and government purchasing. Those levers are genuinely fast, and they are the ones that have actually moved the industry so far: we saw it when export restrictions turned a single frontier model into a geopolitical asset. They are also opaque, largely unreviewable, and entirely reversible by the next administration. A czar is not an institution.

There is a second gap, and it is the one we keep returning to. A task force without a measurement apparatus is a press release with a deadline. The hard part of AI governance is not declaring leadership; it is knowing, in numbers, what these systems can do that last year's could not. Most public benchmarks are saturated to the point of uselessness — the ones that still discriminate are expert, unguided tasks in domains like cyber-offense, and they are run by a handful of institutes with tiny budgets. If the 120-day report comes back with capability thresholds, evaluation funding and a definition of what would trigger a response, it will be a serious document. If it comes back with the word "lead" in every paragraph, it will be a slogan.

The long view stays intact, and it is why this matters at all. The destination of this technology — curing disease, extending healthy life, producing enough that people can work on what they love — is worth governing well, which means governing on evidence rather than on vibes or panic. Naming a desk for AI inside the US government is, in itself, progress; for two years there was no one to call. But a safety architecture resting on voluntary pledges, personal proximity to the president and a 120-day memo is fragile by construction. Watch the report, not the appointment.

🔗 Related on Zendoric

Sources & references