AI Washing: SEC Sues $1.2 Billion Fund Manager, Alleging the 'Proprietary AI Engine' Was Two Analysts and a Spreadsheet

NEW YORK — The Securities and Exchange Commission sued a $1.2 billion hedge fund manager on Wednesday, alleging that the "proprietary AI engine" it told investors was selecting its trades did not exist, and that the trades were chosen by two analysts working from a commercially available stock screener and a spreadsheet. The complaint against Ardent Quant Capital, filed in federal court in Manhattan, is the largest AI-washing case the agency has brought and one of the few it has chosen to litigate.

"Investors were told they were buying a machine. They were buying two people with opinions," the agency's enforcement director said in a statement announcing the case.

The allegations

According to the complaint, Ardent raised about $400 million between 2024 and 2025 for a fund whose pitch materials, website and Form ADV brochure described a "proprietary machine-learning engine trained on two decades of alternative data" that "autonomously identifies and sizes positions with no human discretion." The fund charged a two percent management fee and a twenty percent performance fee, which its materials justified by reference to the engine's "structural edge." The SEC alleges that the engine was a licensed screening tool configured with a dozen conventional factors; that every position was chosen, sized and timed by the firm's founder and one analyst; that an internal email described the AI language as "what allocators want to hear"; and that when a large pension investor's due-diligence team asked for a demonstration, the firm showed a dashboard built for the meeting. The fund's returns, the complaint says, were unremarkable, which investors might have accepted at ordinary fees but did not know they were paying premium fees for.

The charges

• Advisers Act. Sections 206(1) and 206(2), the general anti-fraud provisions for investment advisers, and Section 206(4) with Rule 206(4)-8, which prohibits advisers from making untrue statements of material fact to investors in pooled vehicles. Because 206(2) does not require scienter, the SEC has a path to liability even if it cannot prove intent.
• Marketing Rule. Rule 206(4)-1 prohibits advertisements that include untrue statements of material fact or statements the adviser cannot substantiate on demand. The complaint alleges Ardent could not substantiate the engine's existence, let alone its training.
• Securities Act and Exchange Act. Section 17(a) and Section 10(b) with Rule 10b-5, for the fund interests sold on the basis of the claims.
• Relief. Permanent injunctions, disgorgement of management and performance fees attributable to the AI representations, civil penalties, and an order barring the founder from the industry.

Ardent said in a statement that its disclosures "accurately described a systematic process that uses machine-learning tools" and that the SEC was "second-guessing a marketing vocabulary the entire industry uses." It said it would contest the case.

Where the case fits

The SEC coined the term "AI washing" in 2023 and brought its first cases in March 2024, settling with two advisers, Delphia and Global Predictions, for a combined $400,000 over claims of AI capabilities they did not have. A settlement with Presto Automation over its AI voice product followed in January 2025, and in April 2025 the SEC and prosecutors charged the founder of the shopping app Nate, whose "AI-powered" checkout was performed by contractors. The agency's leadership changed in 2025 and its appetite for novel theories diminished, but it has continued to describe AI washing as plain fraud, and its examination priorities for the past two years have listed AI representations by advisers. The Ardent case differs from its predecessors in scale, in the sophistication of the investors involved — pension funds and endowments rather than retail clients — and in the firm's decision to fight.

Investors were told they were buying a machine. They were buying two people with opinions.

SEC enforcement director, in a statement

What the case will test

Ardent's defense previews the questions a court will have to answer. Where is the line between a "machine-learning engine" and a screening tool with factor weights? Is a claim of "no human discretion" false if humans override the tool only occasionally, or only if they override it always? And is AI vocabulary material to sophisticated investors who conducted their own diligence? The SEC's answer is that the firm's own email settles the first two questions and the fee premium settles the third; the firm's answer is that "AI" has no fixed meaning and that investors got the strategy the documents described. A ruling on a motion to dismiss, expected next year, will be the first judicial word on the theory.

What it means for advisers

• Substantiate. The Marketing Rule requires a reasonable basis for every material claim and the ability to demonstrate it on demand. For AI claims, that means documentation of what the model is, what it does and how often humans intervene.
• Be consistent. Pitch decks, Form ADV, websites and diligence responses are read together; the complaint quotes all four.
• Mind the fee link. Where fees are justified by a technological edge, the edge is material by definition.
• Review internal communications. The email about "what allocators want to hear" is the complaint's centerpiece.

The Counsel will report the firm's answer and any motion to dismiss. Investors in the fund have not been told to take any action; the SEC's complaint seeks disgorgement into a fund for their benefit if it prevails.

This report is for general information only and is not legal advice. Readers should consult a licensed attorney about their own circumstances. Docket numbers and citations are to the official reporters unless otherwise stated.