A company may describe a product, investment strategy, or service as “AI-powered.” That description is not unlawful merely because it is promotional. The legal problem begins when a concrete and material statement about the technology, automation, data, performance, customers, or financial results is false—or when important facts are omitted and the remaining statement becomes misleading.

These disputes are often called AI washing cases. However, AI washing is a description, not a separate cause of action. An investor, customer, or business still must identify an applicable legal claim and connect the challenged statement to a transaction and legally recognized loss.

Key points include:

A vague claim that a company is “innovative” may be nonactionable puffery, while a measurable representation about automation, accuracy, proprietary technology, revenue, or human involvement may be materially different.
A private federal securities claim requires more than proof that a statement was wrong. Materiality, scienter, reliance, economic loss, and loss causation may all be disputed.
SEC and FTC enforcement actions do not automatically establish a private right to damages.
The most important evidence often comes from the gap between public claims and internal technical or operational records.
A market decline, business failure, or disappointing product does not by itself prove fraud.

What does “AI washing” mean?

AI washing generally refers to overstating or mischaracterizing how artificial intelligence is used. Examples may include claims that:

a product is proprietary when core technology is supplied by a third party;
a service is automated when people perform much of the work behind the scenes;
a model achieves an accuracy or completion rate that the available testing does not support;
an investment process uses AI or machine learning when it does not use those tools as represented;
AI has produced customers, revenue, cost savings, or competitive advantages that cannot be substantiated; or
known limitations, human intervention, data problems, or material risks are omitted from an otherwise specific statement.

Context matters. Courts evaluate the complete statement, the audience, the information already available, and whether a reasonable investor or buyer would consider the disputed fact important. Predictions and opinions may also require a different analysis from statements about present or historical facts.

When can an AI claim support a securities lawsuit?

For a private claim under section 10(b) of the Securities Exchange Act and Rule 10b-5, the Eleventh Circuit identifies six elements: a material misrepresentation or omission, scienter, a connection with the purchase or sale of a security, reliance, economic loss, and loss causation. Carvelli v. Ocwen Financial Corp., 934 F.3d 1307 (11th Cir. 2019).

That standard makes the exact wording important. Generalized optimism and vague promotional language may be treated as puffery. A statement can become more significant when it makes a verifiable claim about such matters as model ownership, automation rates, human involvement, accuracy, deployment, revenue, or customer adoption. An omission ordinarily requires a duty to disclose or a prior statement that becomes misleading without the omitted information.

Private securities complaints must also satisfy heightened federal pleading rules. The claimant generally must identify the challenged statements, explain why they were misleading when made, and allege facts supporting a strong inference of scienter as to each responsible defendant.

Loss causation is a separate issue. In a public-market case, a claimant often tries to identify information that revealed the previously concealed truth, a related price decline, and facts separating that decline from other market or company-specific causes. The announcement of an investigation, standing alone, does not necessarily reveal that earlier statements were false. Meyer v. Greene, 710 F.3d 1189 (11th Cir. 2013).

The firm’s guide to Securities Fraud: Rule 10b-5 and Interstate Commerce explains the federal framework in greater detail.

What about private-company investments or commercial transactions?

AI misrepresentation is not limited to publicly traded stock. A private offering may still involve a security, although the claim, forum, available presumptions, and measure of damages can differ. In other transactions, the governing documents and facts may support—or defeat—claims for fraudulent misrepresentation, negligent misrepresentation, breach of contract, breach of warranty, or a statutory remedy.

Under Florida law, fraudulent misrepresentation requires a false statement of material fact, knowledge of falsity, an intention to induce another person to act, and injury to a person who acted in reliance on the representation. Butler v. Yusem, 44 So. 3d 102 (Fla. 2010). The transaction documents, disclaimers, integration provisions, allocation of risk, and nature of the alleged statement can all affect the analysis.

Not every regulatory statute creates a private damages action. The legal theory therefore should follow the claimant, the transaction, the defendant, and the remedy actually authorized by law.

Recent AI-misrepresentation enforcement examples

Government actions illustrate the kinds of factual discrepancies regulators examine. They do not establish liability in a separate private case.

Delphia and Global Predictions

In March 2024, the SEC announced settled proceedings against two investment advisers over statements about their purported use of AI. According to the SEC, Delphia represented that it used AI and machine learning incorporating client data in its investment process when it did not have the claimed capabilities. The SEC also found that Global Predictions made false or misleading claims about its AI-driven forecasts and regulatory status. The advisers resolved the matters without admitting or denying the SEC’s findings and agreed to combined civil penalties of $400,000. See the SEC’s March 18, 2024 announcement.

Presto Automation

In January 2025, the SEC announced settled charges against Presto Automation concerning statements about its drive-through ordering product. The SEC found that certain disclosures did not adequately explain third-party technology and human intervention and that reported automation terminology created a misleading impression. The order found violations of section 17(a)(2) of the Securities Act and reporting and controls provisions. No civil penalty was imposed because of the company’s cooperation and remediation. See the SEC’s Presto proceeding.

Nate

In April 2025, the SEC filed a civil complaint against the founder and former chief executive of Nate, Inc. The complaint alleges that investors were told an e-commerce application used AI to complete purchases without human involvement, while contract workers allegedly performed much of the process manually. The SEC alleges that more than $42 million was raised through stock sales. On June 23, 2026, the district court stayed the SEC action pending the parallel criminal case. These remain allegations, not findings after trial. See the SEC litigation release concerning Nate; SEC v. Mantinan, No. 1:25-cv-02937, ECF No. 32 (S.D.N.Y. June 23, 2026).

The SEC later identified fraud involving AI and machine learning as a priority of its Cyber and Emerging Technologies Unit. Its fiscal-year 2025 enforcement report again identified the Nate action as an AI-related investor-protection matter. See the SEC’s fiscal-year 2025 enforcement results.

Deceptive AI claims outside securities transactions

AI claims can also create consumer or commercial issues outside the securities laws. In August 2026, the FTC finalized consent orders concerning representations that an “active listening” advertising service used AI and consumer voice data as described. The orders resolved the FTC’s allegations and imposed monetary and conduct requirements. See the FTC’s August 27, 2026 announcement.

These examples show why the applicable statute and procedural posture must be stated accurately. A settlement, agency finding, pending civil complaint, and criminal indictment do not carry the same legal meaning.

Evidence that may matter in an AI-misrepresentation dispute

A useful investigation compares what was represented with what the technology and business actually did during the relevant period. Depending on the matter, the evidence may include:

pitch decks, offering documents, subscription agreements, SEC filings, websites, demonstrations, and sales presentations;
emails, text messages, recorded calls, investor updates, and board materials;
source-control records, model documentation, product requirements, testing protocols, accuracy reports, and incident logs;
contracts with AI vendors, data providers, contractors, and human-review services;
records showing when the company obtained, deployed, or discontinued relevant technology;
evidence of human intervention, exception handling, failed transactions, or manual processing;
customer, revenue, usage, and performance data supporting or contradicting public claims;
documents showing who approved the statement and what that person knew at the time;
the investor’s decision documents and communications; and
evidence connecting disclosure of the truth to the alleged loss while accounting for other causes.

The inquiry should focus on contemporaneous evidence. A product’s later failure does not prove that every earlier projection was fraudulent. Conversely, a polished demonstration does not resolve whether the company accurately described the process operating behind it.

Common defenses and limitations

Potential defenses depend on the claim and record. They may include:

the statement was true when made;
the challenged language was immaterial puffery, opinion, or a protected forward-looking statement;
meaningful cautionary information disclosed the relevant risk;
there was no duty to disclose the omitted information;
the defendant had a reasonable factual basis and lacked fraudulent intent or severe recklessness;
the claimant did not rely on the statement or the transaction documents contradict the asserted reliance;
the loss resulted from market conditions, execution problems, competition, or another disclosed risk rather than the alleged misrepresentation;
the instrument or transaction does not support the asserted securities claim; or
the requested remedy is unavailable under the statute invoked.

An SEC or FTC inquiry may be significant, but the existence of an investigation alone is not proof that a private plaintiff can establish liability or loss causation.

Practical steps after suspected AI misrepresentation

An investor or business confronting a disputed AI claim should consider the following steps before evidence disappears or positions harden:

Preserve the exact representation, including its date, speaker, audience, and surrounding context.
Preserve transaction documents, communications, demonstrations, account records, and versions of relevant webpages or presentations.
Identify what decision was made because of the representation and when that decision occurred.
Compare the representation with technical, vendor, staffing, testing, and operational records from the same period.
Separate the alleged misrepresentation from ordinary product risk, market movement, and subsequent events.
Calculate the claimed loss and identify the evidence connecting it to disclosure or materialization of the concealed fact.
Review arbitration clauses, forum provisions, limitations periods, notice requirements, and available remedies before filing suit or making public accusations.

Reporting suspected misconduct to a regulator and pursuing a private recovery are different processes. Each requires its own jurisdictional, procedural, and remedial analysis.

Evaluating an AI-misrepresentation claim or defense

Gherman Legal represents businesses, investors, financial professionals, and other parties in fraud, investment, and commercial disputes in Miami and throughout Florida. An early assessment can identify the exact statement, applicable transaction documents, potential claims and defenses, evidence-preservation needs, forum, and realistic measure of loss.

Learn more about the firm’s Fraud, Finance & Securities Litigation practice and its selected case results.

Past results depend on the facts and law of each matter and do not guarantee or predict a similar result. This article provides general information and is not legal advice. Reading it or contacting the firm does not create an attorney-client relationship.

AI Misrepresentation Lawsuits: Claims, Proof, and Defenses