AI Washing PR: 4 Companies Regulators Challenged

Four companies drew regulator action over artificial intelligence (AI) claims: Delphia said it used client data in AI to pick winning investments, Presto said its drive-thru AI removed the need for human order-takers, Nate said its shopping app completed purchases without human help, and DoNotPay called itself the world's first robot lawyer. Each claim outran the technology. Here is what each said, what regulators found, and how a tech or fintech brand can check its own AI claims first.

By Kyle Porter. Last updated October 22, 2026.

What is in this article?

• What are the four claims at a glance?

• What did Delphia claim about client data and AI?

• What did Presto claim about its drive-thru AI?

• What did Nate claim about automated shopping?

• What did DoNotPay claim with "robot lawyer"?

• What do the four AI cases have in common?

• How can a tech brand check an AI claim before it runs?

• Frequently asked questions

What are the four claims at a glance?

The four claims were made to investors or consumers, were challenged by US securities and consumer regulators between 2024 and 2025, and each came down to the same gap between the pitch and the product.

CompanyAI claimRegulatorDateOutcome

DelphiaUsed client data in AI to pick investmentsSecurities and Exchange Commission (SEC)March 2024$225,000 civil penalty Presto AutomationOwn AI removed the need for human order-takingSECJanuary 2025Settled order, no civil penalty NateApp used AI to complete purchases without humansSEC and Department of Justice (DOJ)April 2025Civil and criminal charges (allegations) DoNotPay"The world's first robot lawyer"Federal Trade Commission (FTC)2024 to 2025$193,000 order, claims restricted

What did Delphia claim about client data and AI?

Delphia told clients it fed their data into AI to predict which companies would succeed, and the SEC said the firm never built that capability. The SEC fined Delphia $225,000 in March 2024 in one of its first AI washing settlements. According to reporting on the settlement, the firm claimed its AI could "predict which companies and trends are about to make it big," and the SEC said that claim was false because the firm did not have those capabilities.

The timeline is the uncomfortable part. A legal analysis of the order notes that Delphia stated from 2019 through 2023 that it used client data in AI and machine learning algorithms, when it had neither used client data nor built such an algorithm, as the firm itself acknowledged in a 2021 SEC examination. The claim stayed in marketing for two years after that.

The same day, the SEC settled with a second adviser, Global Predictions, for $175,000. The agency said Global Predictions falsely called itself the "first regulated AI financial advisor". Neither firm admitted or denied the findings. SEC Chair Gary Gensler, the official who led the agency at the time, said AI washing hurts investors, and the SEC's enforcement director said firms that claim AI use must make sure the claim is accurate.

What did Presto claim about its drive-thru AI?

Presto Automation said its own AI product eliminated the need for human order-taking, and the SEC said the vast majority of orders still needed a person. On January 14, 2025, the SEC announced settled charges against the restaurant-technology company over false and misleading statements about its flagship product, Presto Voice, made from November 2021 through May 2023.

The SEC's order described two problems. First, Presto did not disclose that, for a period, the speech recognition technology in every deployed unit was owned and operated by a third party. Second, after Presto deployed its own technology with some customers, it falsely claimed that product eliminated human order-taking, when most drive-thru orders required human intervention.

This was the first AI washing action against a company that had been publicly listed. The D&O Diary noted that no civil penalty was imposed. Presto had left Nasdaq in September 2024, before the order. The cost was not the penalty. It was the public record that the headline product leaned on human workers.

What did Nate claim about automated shopping?

Nate's founder told investors the app used AI to complete online purchases without human help, and prosecutors allege contract workers did the work by hand. On April 9, 2025, the SEC and federal prosecutors in New York filed parallel civil and criminal actions against founder and chief executive Albert Saniger.

The numbers in the allegations are specific. According to a law firm summary of the filings, Saniger raised more than $42 million from investors by claiming the app used AI, machine learning and neural networks to process transactions, when virtually all purchases were completed manually by contract workers in the Philippines, Romania and elsewhere. A summary of the charges says investors were told automation rates ran between 93 and 97 percent, and the SEC alleged the real rate was essentially zero.

These are allegations, and Saniger is entitled to defend against them. For communications teams, the shift in risk is the point: this was when AI washing moved from civil penalties to a criminal wire fraud count.

What did DoNotPay claim with "robot lawyer"?

DoNotPay marketed its subscription service as "the world's first robot lawyer," and the FTC said the company never tested whether the AI performed at the level of a human lawyer. The FTC announced its complaint in September 2024. According to the FTC's announcement of the final order, DoNotPay did not test whether its "AI lawyer" operated at human-lawyer level and did not hire or retain attorneys to check the quality of its legal features.

The final order, approved 5 to 0 on January 16, 2025, requires DoNotPay to pay $193,000 in monetary relief and notify subscribers from 2021 through 2023. It also bars the company from advertising that its service performs like a real lawyer unless it has evidence to support the claim.

This case is the consumer-facing version of the same problem. No one alleged that the chatbot did nothing. The FTC's point was that a comparison to a human professional is a testable claim, and DoNotPay had not run the test.

What do the four AI cases have in common?

All four companies described the AI as more capable, more autonomous or more tested than it was, and all four claims were checkable. The pattern:

• Delphia said client data powered its AI, but no such algorithm existed.

• Presto said human order-taking was eliminated, but people still handled most orders.

• Nate said purchases were automated, but humans completed the transactions.

• DoNotPay said its AI performed at lawyer level, but it never tested that.

Two of the four involved hidden human labor. A claim like "fully automated" or "AI-powered" implies what is happening behind the interface, and reporters and regulators can find out with a single filing or a single former employee.

The same pattern shows up in climate marketing, where regulators challenge claims that describe one part of a product as if it described the whole. AI washing is the technology-sector version of that problem.

For related reading, see the technology PR agency overview, the AI PR agency page, and a review of fintech PR campaigns.

How can a tech brand check an AI claim before it runs?

Test each AI claim against what the system does today, not what the roadmap says it will do. A short pre-launch check:

• List each task the AI performs, then mark whether software, a person or a third party does it.

• Replace "fully automated" with a measured rate. If humans review any share of the work, state the measured share and the date.

• Name third-party technology internally. If another company's model powers the feature, know that before any press release says "our AI."

• Do not compare the product to a human professional without a test. "Lawyer-level" or "doctor-level" needs documented testing against that standard.

• Check investor decks and press releases against each other. The SEC cases turned on statements to investors, not just ads.

• Keep a dated evidence file. For every AI claim, store the test, the date and the person who approved it.

What should a tech brand take from these cases?

Delphia, Presto, Nate and DoNotPay each learned that an AI claim is a factual claim, and regulators treat it that way. The brands that stay out of trouble describe what the system does today, say where people are involved, and keep the proof on file. Before any launch, funding announcement or press release, each AI statement should be matched to a test result or a named internal owner who can produce one.

Frequently asked questions

What is AI washing?

AI washing is overstating or misrepresenting how a product or company uses artificial intelligence. It includes claiming AI capabilities that do not exist, hiding that people or third parties do the work, and citing performance figures that cannot be supported.

Which regulators have taken action on AI washing?

In the cases above, the US Securities and Exchange Commission acted against Delphia, Global Predictions, Presto Automation and Nate's founder, the Department of Justice brought parallel criminal charges in the Nate matter, and the Federal Trade Commission acted against DoNotPay.

Was anyone fined for AI washing?

Yes. The SEC fined Delphia $225,000 and Global Predictions $175,000 in March 2024, and the FTC's final order required DoNotPay to pay $193,000. Presto's settled order carried no civil penalty.

Is it illegal to say a product is "AI-powered"?

Not by itself. The cases turn on whether the claim is true and supported. A claim that overstates what the AI does, hides human involvement or lacks testing can be treated as misleading under securities or consumer protection law.

Can PR teams be exposed over AI claims?

Communications teams write the statements regulators examine, including press releases, websites, investor decks and social posts. Several of the cases above rested on public marketing materials, so PR review of AI claims is a practical safeguard.

How do I describe AI honestly in marketing?

State what the system does today, say where humans are involved, give measured figures with dates, and avoid comparing the product to a human professional unless you have documented testing.