The SEC's AI Washing Enforcement Actions, Dated
9 min read · updated August 11, 2026
“AI washing” is not a statutory term and there is no AI disclosure rule. What the Securities and Exchange Commission has done is charge ordinary misrepresentation, where the thing misrepresented happened to be artificial intelligence. The distinction matters, because it tells you exactly which conduct is exposed.
What AI washing means as a charge
The Commission has been explicit that there is nothing special about the subject matter. The prohibitions used are the general antifraud and advertising provisions of the federal securities laws, applied to statements about AI capability in the same way they would be applied to statements about assets under management or investment process.
That produces a clean test. The exposure is not created by using AI, nor by describing it, but by a statement about AI capability that is materially false or misleading and that a reasonable investor or client would consider important. Three recurring fact patterns appear in the orders: claiming to use AI where none is used at all; describing a capability in the present tense that is aspirational or under development; and describing AI as central to a process where it plays a marginal role. The third is the one honest firms fall into.
March 2024: the first two settled orders
On 18 March 2024 the Commission announced settled charges against two investment advisers, Delphia (USA) Inc. and Global Predictions Inc., for making false and misleading statements about their use of artificial intelligence. The Commission published a press release describing both. SEC press release, 18 March 2024.
Delphia (USA) Inc.
The Commission’s order found that the firm made statements in filings, in a press release and on its website claiming that it used artificial intelligence and machine learning applied to client data to predict which companies and trends would grow. The order found the firm did not have the capability it described. The order imposed a civil penalty of 225,000 dollars, alongside a censure and a cease-and-desist order.
Global Predictions Inc.
The Commission’s order found that the firm made false and misleading claims on its website and on social media, including describing itself as the first regulated AI financial adviser, and making claims about AI-driven forecasts. The order imposed a civil penalty of 175,000 dollars, again with a censure and a cease-and-desist order. The order also addressed a hedge clause and compliance deficiencies, which is a useful reminder that these examinations rarely find only one thing.
Both were settled administrative proceedings in which the firms neither admitted nor denied the findings. That is the standard posture and it means the orders are not adjudications of contested fact; they are findings the respondent agreed not to contest.
Later matters
The March 2024 orders were the beginning rather than the whole record. The Commission subsequently brought further matters in this area, and they divide into two kinds that should not be described interchangeably.
Settled administrative proceedings against advisers. In October 2024 the Commission announced settled charges against Rimar Capital USA, Inc., a related entity and its principal, concerning claims about AI-driven trading technology. As with the March matters, the respondents settled without admitting or denying the findings. Read the order for the specific findings and the penalties.
Contested actions against issuers and individuals. In 2024 the Commission charged the founder of a recruitment technology company, Joonko, in federal district court with defrauding investors through claims about the company’s AI-driven capabilities and customer base; the Department of Justice announced a parallel criminal matter. In January 2025 the Commission announced settled charges against Presto Automation Inc. concerning statements about the AI capabilities of a drive-through ordering product. These are securities fraud and reporting matters rather than Advisers Act matters, and the applicable provisions differ accordingly.
The Commission also published an investor alert on artificial intelligence and investment fraud, which is educational material rather than enforcement. SEC Office of Investor Education and Advocacy.
The provisions the SEC uses
Knowing which provision applies to you is more useful than the list of names, because the elements differ.
| Provision | Description |
|---|---|
| Advisers Act 206(2) | Prohibits an adviser from engaging in a transaction, practice or course of business that operates as a fraud or deceit on a client or prospective client. Notably, it does not require proof of scienter — negligence suffices, which is why it appears in settled orders against advisers who may not have intended to mislead. |
| Advisers Act 206(4) and Rule 206(4)-1 | The Marketing Rule. It prohibits untrue statements of material fact and unsubstantiated material statements in an adviser’s advertisements, and requires a reasonable basis for believing material statements can be substantiated on demand. 17 CFR 275.206(4)-1. |
| Advisers Act 206(4)-7 | The compliance rule. Requires written policies and procedures reasonably designed to prevent violations. An adviser making AI claims without a process for reviewing them is exposed here independently of whether any claim was false. |
| Advisers Act 207 | Prohibits materially false or misleading statements in registration filings, including Form ADV. AI capability claims frequently appear in Form ADV brochures. |
| Securities Act 17(a) and Exchange Act 10(b) / Rule 10b-5 | The general antifraud provisions, used against issuers and individuals raising capital on AI claims. These require scienter for the 10(b) route and are typically litigated in federal district court rather than settled administratively. |
The Marketing Rule’s substantiation requirement is the practical centre of gravity for advisers. It is not enough that a capability claim is arguable; the adviser must have a reasonable basis for believing it can substantiate the claim on demand, and the Commission has asked for that basis. An adviser that cannot produce documentation of what its system does has a Marketing Rule problem before anyone reaches the question of whether the claim was false.
Reading the record accurately
Four cautions, each of which corrects a claim that circulates.
- There is no AI rule. No disclosure rule requires a firm to describe its AI use, and none prohibits AI marketing. Every matter to date is an application of existing antifraud and advertising provisions.
- Settled orders are not precedent. A respondent neither admitting nor denying findings has not litigated anything. The orders show what the staff considered chargeable, which is useful and is not the same as law.
- Penalty figures should come from the order. Secondary sources routinely conflate a penalty with disgorgement, or attribute a combined figure to one respondent. The orders are published on the Commission’s site and state the amounts.
- Enforcement priorities move with the Commission.Leadership changes alter what is charged and how often, and 2025 brought significant change in emphasis at the agency. What does not change without rulemaking or legislation is the underlying prohibition: the Marketing Rule and the antifraud provisions are not AI policy and were not adjusted.
The parallel exposure under the Federal Trade Commission’s general prohibition on deceptive practices reaches firms that are not advisers at all — see AI washing under FTC Act section 5 and the law on AI marketing claims for the broader picture, including the substantiation standard that applies outside the securities context.