What is AI washing, and who has actually been charged over it?

What "AI washing" actually means

"AI washing" is not a law. It is a market term, borrowed from "greenwashing," for a company exaggerating, faking or simply not disclosing the real limits of its artificial intelligence, usually to look more fundable or more cutting-edge than it actually is. No regulator has a statute with that name on the books. Instead, the Securities and Exchange Commission and the Federal Trade Commission have both found a faster route: treat an inflated AI claim the same way they would treat any other false statement, and charge it under securities law or ordinary consumer-protection law.

That distinction matters because it means the bar is not some special AI standard. It is the same test regulators apply to any product claim: is it true, can you back it up, and did you tell investors or customers the parts that would change their decision. A company that says "AI-powered" when the work is mostly done by people, or that claims a capability it never tested, is exposed exactly the way a company would be for lying about revenue or safety testing.

The SEC's first move: investment advisers caught bluffing

The SEC's opening AI washing cases, announced on 18 March 2024, targeted two investment advisers. Delphia (USA) Inc., a Toronto-based firm, had claimed from 2019 to 2023 that it "put[s] collective data to work to make our artificial intelligence smarter," a statement the SEC found was false. San Francisco-based Global Predictions Inc. had called itself the "first regulated AI financial advisor" and misrepresented what its platform's forecasts could do.

Both firms settled without admitting or denying the findings. Delphia paid a $225,000 civil penalty and Global Predictions paid $175,000; both were censured and ordered to stop. Then-SEC Chair Gary Gensler put the standard plainly: "Investment advisers should not mislead the public by saying they are using an AI model when they are not." The cases were brought under the SEC's Marketing Rule and general anti-fraud provisions, not any AI-specific rule, because none existed or exists now.

The FTC goes after fake AI schemes

Six months later, on 25 September 2024, the FTC announced Operation AI Comply, charging five companies in a single sweep. DoNotPay had marketed itself as the "world's first robot lawyer" without testing whether its output matched a real lawyer's work and without employing any attorneys. Ascend Ecom and FBA Machine both allegedly promised consumers guaranteed income from AI-powered online storefronts; the FTC says Ascend Ecom's scheme cost consumers more than $25 million and FBA Machine's cost more than $15.9 million. Ecommerce Empire Builders made similar unsupported income claims, and Rytr sold an AI writing tool the FTC says generated detailed, likely false consumer reviews on demand.

Then-FTC Chair Lina Khan did not mince words: "The FTC's enforcement actions make clear that there is no AI exemption from the laws on the books." The agency's own guidance adds a useful line for anyone sizing up a vendor's pitch: using an AI tool somewhere in development does not, by itself, make the finished product "AI-powered."

When a public company got caught: Presto's drive-thru "AI"

The cases above involved advisers and smaller operators. In January 2025, the SEC charged its first public company over AI washing: Presto Automation Inc. Presto marketed Presto Voice as AI that could take drive-thru orders without a human. The SEC found that for a period of time, the product was largely staffed by human order-takers the company hired, trained and supervised in the Philippines and India, who processed the vast majority of orders placed through the system. The SEC also found Presto had failed to disclose that some of the underlying technology was owned and operated by a third party rather than built in-house.

Presto settled without admitting or denying the findings and consented to a cease-and-desist order. The SEC did not impose a civil penalty, citing the company's financial condition and its cooperation. Three months later, in April 2025, the SEC and the Department of Justice brought a related kind of case against an individual: Albert Saniger, founder and former CEO of the shopping app Nate, Inc. The SEC alleges Saniger told Seed and Series A investors that Nate's app used AI to complete purchases "without human involvement," after the company had raised more than $42 million. In fact, the SEC says, contract workers manually typed in a large share of the orders, and the app's real success rate fell short of what investors were told. The DOJ filed a parallel criminal case the same week, adding a wire fraud charge on top of the securities fraud allegations.

How to actually tell the difference

None of this means every "AI-powered" label is a lie, most are not. But the pattern across these cases is consistent: regulators ask what model is actually running, who built it, how much of the output is genuinely automated, and whether the company tested the claim before making it. The same questions work for a reader sizing up a product. Ask what happens when you push past the demo, whether the company names the underlying model, and whether independent benchmarks or user reports back the claim up. Plenty of tools marketed as one company's breakthrough AI turn out to be, as an AI Newspedia explainer on AI wrappers lays out, a thin layer built on someone else's model, which is not illegal on its own but becomes a problem the moment the marketing implies otherwise.

For founders and product teams, the lesson from Presto and Nate is specific: a claim that a human safety net does not exist, or that a capability was tested, is the kind of statement that turns ordinary marketing exaggeration into a securities or consumer-protection violation. For buyers, the lesson is just as specific: a vendor that will not say which model it uses, or what a human reviews before the output reaches you, is the pattern every one of these cases eventually ran into.

Our take

Calling your product "AI-powered" is not illegal, and neither regulator has criminalized the phrase itself. What both the SEC and FTC have shown, in cases against two investment advisers, five scheme operators, a public company and a startup founder, is that they will treat a false or untested AI claim exactly like any other false statement, with real fines, cease-and-desist orders and in one case a criminal charge. The safest reading for anyone selling or buying "AI" right now: the claim has to survive the same scrutiny as any other business promise, because regulators are already checking.