FTC Opens Unfair-Practices Inquiry Into OpenAI and Anthropic

The Federal Trade Commission is investigating OpenAI and Anthropic over potential consumer harms, examining whether the two artificial intelligence labs violated federal laws that prohibit unfair and deceptive business practices.

The Federal Trade Commission has opened an investigation into OpenAI and Anthropic, the two best-funded developers of consumer-facing artificial intelligence assistants, over potential consumer harms. According to NYT Technology, the agency will examine whether the A.I. labs have broken federal laws that prohibit companies from engaging in unfair or deceptive practices.

That framing matters more than it may first appear. The inquiry is not, on its face, an antitrust case about market structure, model licensing or cloud partnerships. It is a consumer-protection matter — the branch of the FTC's authority that governs what a company tells the public about its product and how that product treats the people using it.

What the FTC's unfair-and-deceptive standard actually covers

The prohibition on "unfair or deceptive acts or practices" is the FTC's broadest and most flexible tool. Deception turns on representations: what a company claims, implies or omits in a way likely to mislead a reasonable consumer about something material to their decision. Unfairness is a separate test, aimed at conduct that causes substantial injury consumers cannot reasonably avoid and that is not outweighed by offsetting benefits.

Applied to chatbots, that standard can reach a long list of things without the agency ever having to argue that either company is too big. Among the categories the authority plausibly touches:

• Safety and capability claims. Public statements about guardrails, refusal behavior, accuracy or suitability for sensitive uses are marketing representations. If a product does not perform as advertised, that is classic deception territory.
• Harms to vulnerable users. The unfairness prong does not require a false statement at all — only injury that users could not reasonably sidestep. That is the doctrinal hook for concerns about minors, mental-health interactions and emotionally manipulative design.
• Data handling. What is collected from conversations, whether it trains future models, how consent is obtained and whether deletion promises are honored have been staple FTC theories in other industries for two decades.
• Subscription and billing mechanics. Enrollment flows, auto-renewal disclosure and cancellation friction are standard consumer-protection ground for any paid consumer service.
• Dark patterns and defaults. Interface choices that steer users toward disclosing more or paying more than they intended have been treated as actionable design, not mere styling.

None of that means the agency has concluded anything. An investigation is a fact-gathering stage; it can close quietly, end in a consent order with conduct commitments, or escalate into litigation. But the choice of legal theory sets the shape of what comes next — and consumer-protection cases tend to resolve in behavioral remedies that reshape product design, disclosure and default settings rather than in structural breakups.

Why two private companies still register with public-market investors

Neither OpenAI nor Anthropic is listed, so there is no share price to move on the news. The transmission mechanism runs through their commercial partners and customers instead. Both labs sit at the center of a web of cloud contracts, chip orders, enterprise deployments and distribution deals with publicly traded firms, and the economics of those relationships assume that consumer A.I. assistants keep scaling as a mass-market product.

A consumer-protection action that forces changes to onboarding, age assurance, data retention or subscription flows would not destroy that assumption. It would tax it — adding compliance layers, slowing feature releases and, at the margin, trimming conversion rates on paid tiers. For investors trying to underwrite A.I. revenue several years out, the relevant question is not whether a fine arrives but whether the product surface becomes materially harder to iterate on.

The broader tape was not treating the news as a systemic event. As of the last trade at 17:37 GMT on September 30, 2026, the Nasdaq 100 tracker (NASDAQ: QQQ) was at $742.82, up 0.66% from the prior close of $737.93, with a day range of $739.76 to $745.10. The S&P 500 proxy (NYSEARCA: SPY) stood at $767.20, up 0.39% against a previous close of $764.20. The Dow 30 fund (NYSEARCA: DIA) was the laggard at $511.55, down 0.26%. Tech-heavy exposure, in other words, was leading on the session rather than flinching.

The regulatory pattern the inquiry fits into

Consumer protection has become the path of least resistance for regulators approaching A.I. Antitrust cases against the labs would require defining markets that barely existed a few years ago and proving harm in a sector where prices to users have generally been falling. Privacy legislation aimed specifically at generative models has not arrived at the federal level. What exists already, and requires no new statute, is the authority to police what companies say and how their products treat the public.

That gives the FTC a durable entry point regardless of how the technology evolves. It also means the labs are being measured by a standard built for advertisers, subscription services and data brokers — a standard that asks whether a reasonable consumer was misled or injured, not whether the underlying engineering was state of the art.

For OpenAI and Anthropic, the reputational stakes cut differently. Anthropic has built much of its positioning around safety; OpenAI has the larger consumer footprint and therefore the larger exposed surface. An inquiry that questions safety representations lands hardest on whichever company has leaned hardest on them.

What to watch from here

Several markers will indicate whether this stays a paperwork exercise or becomes a defining constraint on consumer A.I.

• Scope disclosure. Whether the investigation is framed narrowly around specific claims or broadly across product design will signal the agency's ambition.
• Company responses. Voluntary product changes announced during an open inquiry usually presage a negotiated settlement.
• Partner language. Watch for new regulatory risk factors in the filings of listed cloud providers, chipmakers and software firms that route revenue through these models.
• State attorneys general. Federal consumer-protection matters frequently attract parallel state actions built on similar unfair-practices statutes.
• Enterprise contracting. Corporate buyers reading indemnity clauses closely could slow procurement, which is where the financial effect would actually show up.

The immediate practical consequence is narrower than the headline suggests: two private companies are answering questions from a regulator using authority it has exercised for generations. The longer-term consequence is that the terms on which consumer A.I. can be marketed in the United States are now being set by a consumer-protection agency, not by the labs themselves.

Key facts

• Companies under investigation: OpenAI and Anthropic (both privately held)
• Legal authority: Federal prohibition on unfair and deceptive practices
• Nasdaq 100 (QQQ): $742.82, +0.66%, as of 17:37 GMT Sept 30, 2026
• S&P 500 (SPY): $767.20, +0.39%, as of 17:37 GMT Sept 30, 2026

Frequently asked questions

What is the FTC investigating at OpenAI and Anthropic?

The Federal Trade Commission has opened an inquiry into potential consumer harms at the two artificial intelligence labs. Specifically, the agency will examine whether the companies have broken federal laws that prohibit businesses from engaging in unfair or deceptive acts or practices. The inquiry is a fact-gathering stage and does not amount to a finding of wrongdoing.

How is this different from an antitrust case?

Antitrust cases concern market power, competition and structure. This inquiry uses the FTC's consumer-protection authority, which asks whether a company misled the public or injured consumers in ways they could not reasonably avoid. Remedies in consumer-protection matters typically involve changes to disclosures, product design and business practices rather than structural separation.

Can I buy shares in OpenAI or Anthropic?

No. Both companies are privately held, so there is no exchange-listed stock and no ticker for either. Public-market investors gain indirect exposure through listed firms that supply computing capacity, semiconductors or distribution to the labs, or that embed their models in commercial software products.

What kinds of conduct could the unfair-and-deceptive standard reach?

The standard can cover marketing and safety claims about a chatbot's behavior, accuracy or guardrails; how conversation data is collected, retained and used for training; consent and deletion practices; subscription enrollment and cancellation mechanics; and interface designs that steer users toward disclosing or spending more than they intended.

How did the market react on the day of the news?

Broad indexes were not signaling alarm. As of the last trade at 17:37 GMT on September 30, 2026, the Nasdaq 100 tracker QQQ was at $742.82, up 0.66%, and the S&P 500 proxy SPY was at $767.20, up 0.39%. The Dow 30 fund DIA was down 0.26% at $511.55.

What happens next in an FTC investigation like this?

Investigations can close without action, resolve through a consent order in which a company agrees to specific conduct changes, or escalate into litigation. Signals to watch include how broadly the agency frames its scope, whether the companies announce voluntary product changes, and whether state attorneys general open parallel cases under similar statutes.

Sources

• F.T.C. Investigates OpenAI and Anthropic Over Potential Consumer Harms — NYT Technology

Photo: StartupStockPhotos · Pixabay Content License — source