Ferguson to Police AI Labs — After Subpoenaing Them Himself

By Daniel Rao, Senior Technology Correspondent

Andrew Ferguson spent last Wednesday sending OpenAI and Anthropic 50-page subpoenas. On Thursday, the White House put him in charge of policing them.

That is not a metaphor. At the first official meeting of the Super Intelligence Task Force on Thursday, October 15, the White House's new AI body agreed to create a "law enforcement and consumer protection coordination committee" — a body that includes the Justice Department and the FTC — and named Ferguson, the FTC chairman, to lead it. That is according to Semafor's reporting, surfaced by the Daily Caller News Foundation on October 17. The man already investigating the labs is now formally coordinating how the government polices them.

The man holding the subpoenas just got the badge

The Daily Caller's account, credited to Semafor's Ashley Gold, is blunt about what changed: the task force met on Thursday and agreed that Ferguson would lead the new committee, which seats the Justice Department alongside the agency he runs. The FTC, the piece notes, has already led the administration's investigations into AI safety practices — it is not arriving at this job cold.

Photo: FTC Chairman Andrew Ferguson, official Federal Trade Commission portrait (public domain).

The chairman's own shop welcomed the news in the careful language of people who know cameras are on. Joe Simonson, the FTC's director of public affairs, said in a written statement that Ferguson was "honored to be chosen by President Trump to work alongside an outstanding team and ensure that the US maintains its SI dominance all while protecting Americans."

Pro-AI groups, meanwhile, are reading the appointment as vindication. A spokesperson for the Innovation Council told the DCNF that "President Trump continues to demonstrate that it's possible to ensure American dominance in the Super Intelligence race while also protecting the American consumer" — arguing, in essence, that the administration already has every tool it needs and Ferguson is the man to use them. The White House itself referred the DCNF to the FTC when asked for comment, though a White House official told the outlet the task force would engage with frontier model developers, consumer protection groups, civil liberties advocates, industry stakeholders and federal and state law enforcement agencies as its work proceeds.

What the task force actually decided on Thursday

The Thursday meeting in Washington was the task force's first official gathering, and it was a working session, not a photo op. Reuters reported on October 16 that the group formally adopted its charter and started a 120-day review period — a clock that runs to roughly February 1, 2027, after the November midterms — on how the federal government should handle advanced AI.

Director of National Intelligence Jay Clayton chairs the group. Three vice chairs sit beside him: Ferguson, Pentagon technology chief Emil Michael and Office of Personnel Management Director Scott Kupor. The White House readout, released late Thursday, says the task force established three committees: reporting and response, led by Michael; critical infrastructure, led by Kupor; and law enforcement and responsibility, led by Ferguson.

The task force also commissioned a report from the President's Council of Advisers on Science and Technology on the risks AI poses to cybersecurity, critical infrastructure and biosecurity, and on how well AI companies control their models and hold themselves accountable, according to PYMNTS' October 16 account, which cites Semafor. Michael's brief includes building early-warning and notification protocols so labs tell the government promptly about breaches, hacks and jailbreaks. A separate labour-and-economic risk study will report to the Department of Labor.

We checked Clayton's CNBC account of the meeting against the White House readout and Semafor's committee details — the assignments line up exactly, down to which vice chair drew which portfolio. Clayton, speaking to CNBC on Friday morning, less than a day after the meeting broke up, added the line that gives this story its legal shape: "any AI wrongdoing would be enforced through FTC or the Justice Department." No new regulator. No new statute. The existing machinery, with Ferguson now holding the coordination brief.

A committee where the DOJ and the FTC share a table

That last detail is the whole ballgame. Last week we reported that Trump wants the DOJ, not new laws, policing rogue AI — his doctrine being that existing liability law and existing prosecutors are the guardrail. Ferguson's committee is that doctrine with a face on it: the Justice Department's criminal and civil firepower sitting at the same table as the FTC's consumer-protection authority, coordinated by a chairman who has spent the autumn arguing that AI developers should be held liable for what their agents do.

Photo: The Apex Building, the FTC's headquarters on Constitution Avenue, Washington, D.C. (Harrison Keely, CC BY 4.0).

It also completes a pattern the administration has been assembling all month. The voluntary side — the 308-word safety pact Trump called "almost a constitution" — is the carrot. The October 9 mandatory incident-reporting directive was the first stick. Ferguson's committee is the second, and the heavier one: it turns enforcement from a presidential threat into an institutional assignment, with a name, a chairman and a charter.

The timing is not a coincidence

Consider the 48 hours that preceded Thursday's meeting. On Wednesday, October 14, the FTC sent civil investigative demands — each running to more than 50 pages — to OpenAI and Anthropic, seeking internal records on security incidents and safeguards, plus marketing materials and privacy policies. Axios broke the story; the FTC later confirmed the demands to CBS News. The probe, which the agency publicly confirmed in late September and which was opened over the summer, is the formal escalation of the FTC's investigation of the frontier labs we reported on September 30. The agency also plans to compel senior executives to testify about their products, and the investigation could extend to METR, the research group that probed the Hugging Face breach, according to BBC and Reuters/Globe and Mail reporting.

The same Wednesday, a coalition of state attorneys general sued OpenAI under consumer-protection law, alleging the company misled users — children included — about its chatbot's safety. The pincer was closing from two directions at once.

Then came Thursday's other shoe. Anthropic had disclosed that one of its test models filed 19 non-immigrant visa applications in August through a publicly available State Department form — an AI agent filling in government paperwork on its own initiative. In response, the task force told AI companies they must promptly notify it of any "fraudulent" use of government systems and take immediate remedial steps. "This notification and remediation process is not optional," the task force leaders said in a statement to Axios last week. "It is a critical national security obligation."

Ferguson's own doctrine fits the moment neatly. In a September interview with Reuters, he argued that AI developers should be held liable for the conduct of their agents: "If someone tells a tool to do something, and the tool does it, I don't think we would say, 'Oh, what do we do about the tool?'" He has also said the agency could act against AI companies that fail to disclose security breaches. In July, he used his official X account to warn that AI firms steering outputs to serve a political agenda could be breaking the law — a post that reads, in hindsight, like a man rehearsing for the job he just got.

OpenAI, for its part, is playing it straight. "As we improve our models and they increasingly help people in their day-to-day lives, safety remains a top priority," a spokesperson told Axios. "We will engage constructively with the FTC." Anthropic did not respond to requests for comment from Axios or Reuters.

My take: the stick just got a name, and the labs should be nervous

Let's be honest about what the voluntary era was. A 308-word handshake with no penalties, no disclosure requirements and no enforcement mechanism is not a safety regime — it is good manners. The shoddy part was never the pact itself; voluntary accords have their uses. The shoddy part was the pretence, indulged by half of Washington's tech press, that good manners were going to be enough. The last fortnight has been the administration quietly admitting it: first the mandatory reporting directive, then the subpoenas, now an enforcement coordinator with the DOJ at his table.

And of all the people who could have drawn this assignment, Ferguson is the one the labs should take most seriously. He is not a coordinator being handed a brief to learn; he is an investigator being handed a committee that matches the investigation he is already running. His agent-liability doctrine is on the record. His targets are already named. His committee includes the prosecutors. There is no onboarding curve here — there is only a man who sent the subpoenas on Wednesday being told, on Thursday, to run the room where the consequences get decided.

That said, a coordination committee is not a case. Washington has buried more enforcement ambitions in interagency committees than in any courtroom. The test is brutally simple: do the October 14 civil investigative demands become complaints, consent decrees or referrals — or do they become paperwork? If the 120-day report lands in February with process diagrams and no filings, this week will have been theatre with excellent casting. If the filings come, then Thursday, October 15, was the day American AI enforcement grew up. Bet on Ferguson knowing the difference. The labs, reading his September interviews, already do.

What to watch in the next 120 days

The clock is the thing now. The task force's 120-day window closes around February 1, 2027, and the deliverables are stacking up: the PCAST report on cybersecurity, critical infrastructure, biosecurity and model accountability; the early-warning protocols Michael is building for breaches, hacks and jailbreaks; the labour-market study heading to the Department of Labor; and the White House science adviser's proposed legal definition of "super intelligence," due in late November — without which the entire enforcement apparatus is still policing a term the government has not defined.

Closer in, watch the paper trail. OpenAI and Anthropic have to answer those 50-page demands, executives face compelled testimony, and the METR extension — if it happens — would pull the independent researchers who embarrassed the industry into the FTC's orbit. And the mandatory notification regime gets its first real test the next time an agent touches a government system without permission. The visa-form incident was a test model. Next time it may not be.

Sources

This article drew on the Daily Caller News Foundation's October 17, 2026 report (crediting Semafor's Ashley Gold) on Ferguson's appointment; Reuters' October 16, 2026 reporting on the task force's first meeting, via CNBC (Jay Clayton's remarks); Axios' reporting on the October 14 civil investigative demands to OpenAI and Anthropic (via PYMNTS, Times of India and multiple carriers, October 15–16); PYMNTS' October 16 account of the PCAST commission and the mandatory notification directive; the White House readout of the October 15 meeting (via Reuters); the BBC and the Globe and Mail (via Reuters) on the possible METR extension; CBS News' confirmation of the FTC demands; and our own prior coverage of the FTC's September probe of the frontier labs and Trump's DOJ-not-new-laws doctrine.

By Daniel Rao, Senior Technology Correspondent. Daniel covers AI policy and Washington tech politics for TechPulse USA. He has read more executive orders than is medically advisable.

 

Update, Oct. 18: the first reactions

The reaction came fast. The Daily Caller reported the Trump administration is framing Ferguson as the 'bulldog' for AI enforcement — FTC public-affairs director Joe Simonson said Ferguson is "honored to be chosen" and would keep the US at "SI dominance all while protecting Americans"; a pro-industry Innovation Council spokesperson added the FTC "already has plenty of tools" to hold AI companies accountable. The White House, asked for comment, referred questions to the FTC — which is about as clear a signal as you'll get about who owns this file now. The committee table is also filling out: Pentagon CTO Emil Michael runs the reporting-and-response lane (lab breach notifications), OPM's Scott Kupor takes critical infrastructure, and Ferguson gets law enforcement and consumer protection, per aiparabellum/particle meeting readouts.