Invest1 publisherNot yet confirmed elsewhere2 min readPublished
FTC staff draft subpoena-like demands for frontier AI companies, a senior official says
The Federal Trade Commission is close to sending civil investigative demands to frontier AI companies, a senior agency official told Semafor. Those demands carry subpoena-like force, and sending them would be the first test of whether the inquiry is more than signalling ahead of the midterms.
The Investor · Invest desk

What happened
- The inquiry surfaced in a New York Post story less than 24 hours after the FTC chair stood with Trump and AI CEOs to announce a nonbinding self-regulation accord.
- According to the official, the demands run to dozens of questions on what the companies have said about their products' dangers and on other consumer harms.
- The official said the agency plans a significant probe and disclosed it early because of public health and safety concerns about AI systems.
- OpenAI's Chris Lehane said the company is still trying to find out more about the investigation, while Anthropic declined to comment.
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- exposure Compelled answers about past statements on product dangers would put the companies' own safety messaging on the FTC's record before the agency decides whether to sue.
- decision Backers of the named companies now face an open federal inquiry that, by Semafor's account, sits well short of any lawsuit or settlement decision.
- contradiction The official says career staff are drafting the demands, but two people told Semafor those staff learned of the probe from the press.
The FTC rarely confirms an investigation before its demands go out, and it is rarer still to do so well before any decision on a lawsuit or settlement [4]. The best-known exceptions, Meta after Cambridge Analytica in 2018 and Equifax after its 2017 breach [8], followed public scandals. A former FTC official explained to Semafor why the agency normally waits. "The simple fact that there is an existing investigation will impact a company's stock price and operations," the official said. [9] By that logic, the companies and their backers have been carrying the cost of this inquiry since the New York Post story [2]. No demand has been sent yet.
William Kovacic, a former Republican FTC chair, said early confirmation is "unusual." It is also "not in any sense improper," he said. [10] "You don't do this unless you want to make clear to the larger public that you're looking at a particular issue," he said [11].
In the first possible outcome, the demands go out at the length the official described and staff work them, so the companies face compelled disclosure on their own safety statements however the matter ends. In the second, the demands go out but the file stalls, and the inquiry amounts to signalling. In the third, the demands never go out at all. Lina Khan, the former FTC chair, expects something close to that. "You have on the one hand, these CEOs of these AI companies wining and dining at the White House," she told ABC. "The idea that the next morning they're going to get some type of credible subpoena from the same administration I think just strains credulity." [19]
I think the first outcome is likelier than Khan allows. The administration's stated approach is to apply the laws on the books without adding new regulatory requirements [12], and Ferguson has said he does not consider the FTC an AI regulator [13]. An agency that has ruled out writing rules has only cases left, and in this probe the demand is the step that comes before any lawsuit or settlement [4]. The counter-case is serious. The administration's 2025 AI Action Plan called for a review of Biden-era FTC investigations to "ensure they do not advance theories of liability that unduly burden AI innovation" [14]. Ferguson has since been appointed to Trump's AI task force [15]. That is an awkward setting for a consumer-harm case against the leading labs. If the demands that arrive run to a handful of questions instead of dozens [5], or reach only METR, the independent evaluator the inquiry may also target [21], Khan was right.
What to watch
- Confirmation from OpenAI or Anthropic that a demand has actually arrived, the first sign the inquiry has moved past the leak.
- Any FTC step under the 2025 AI Action Plan to modify Biden-era orders affecting AI companies while this inquiry is open.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence38
- Adoption
- Insufficient
- Hype gap+25
- Incentives70
- Confidence40
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
The Federal Trade Commission is close to sending investigative demands to frontier AI companies, a senior agency official told Semafor.
- [2]
The FTC's planned AI inquiry emerged in a New York Post story; less than 24 hours before news of the probe, the agency's chair appeared alongside President Trump and AI CEOs to announce a nonbinding accord that pledged industry self-regulation.
- [3]
Civil investigative demands have subpoena-like force.
- [4]
It is rare for the FTC to reveal an investigation before it sends civil investigative demands, and even less typical to do so well before any decision has been made about a potential lawsuit or settlement.
- [5]
The CIDs being sent to AI companies are dozens of questions long, encompassing what the companies have said about the dangers of their products and other questions around consumer harm.
- [6]
Career staff attorneys in the FTC's Bureau of Consumer Protection are drafting the CIDs, the senior FTC official said.
- [7]
The senior FTC official said the agency plans a significant probe and that it was revealed to the media due to public health and safety concerns about AI systems.
- [8]
Exceptions to the FTC's practice of not publicizing investigations include its revelation of an inquiry into Meta's privacy practices after the 2018 Cambridge Analytica scandal, and its 2017 confirmation that it was investigating Equifax after a data breach.
- [9]
A former FTC official said the agency tries not to "publicize investigations in advance, because you don't want people to prejudge the investigation. The simple fact that there is an existing investigation will impact a company's stock price and operations."
- [10]
William Kovacic, a former Republican FTC chair, said confirming an investigation in the press is "unusual," but not without precedent and "not in any sense improper."
- [11]
"You don't do this unless you want to make clear to the larger public that you're looking at a particular issue," Kovacic said.
- [12]
The FTC announcement aligned with Trump advisers' approach to AI safety: apply the laws on the books without adding new regulatory requirements.
- [13]
FTC Chair Andrew Ferguson has said that he does not consider the agency to be an AI regulator.
- [14]
The Trump administration's 2025 AI Action Plan called for a review of all FTC investigations commenced under President Biden to "ensure they do not advance theories of liability that unduly burden AI innovation."
- [15]
Ferguson has since been appointed to Trump's AI task force.
- [16]
"We're trying to find out more about that," OpenAI's chief global affairs officer Chris Lehane said when asked about the investigation.
- [17]
Anthropic, also named in the initial report, declined to comment; METR did not respond to a request for comment.
- [18]
Two people with knowledge of the matter told Semafor that career FTC staff learned about the AI investigation from the press, not agency leadership; the senior official disputed that.
ReportedContestedSource: Two people with knowledge of the matter; senior FTC official, via Semafor2 sources— create a free account to open themView cited source - [19]
"You have on the one hand, these CEOs of these AI companies wining and dining at the White House, the idea that the next morning they're going to get some type of credible subpoena from the same administration I think just strains credulity," former FTC chair Lina Khan told ABC.
ReportedContestedSource: Lina Khan, to ABC, via Semafor2 sources— create a free account to open themView cited source - [20]
The leak raised the question of whether Ferguson plans an in-depth inquiry or a probe designed to convey the appearance of serious action ahead of the midterms.
- [21]
The announcement may also target independent AI evaluator METR.
Sources
1 independent publisher whose own reporting we read for this story.
- semafor.comFTC advancing on its AI safety probe
1 article · October 8, 2026
Topics and entities
Follow any of these and your For You feed starts watching them — no settings page required.