AI Washing in Securities Disclosures: SEC Enforcement Risk for Public Companies

“AI washing” — the practice of overstating a company’s artificial intelligence capabilities in public disclosures, investor materials, and registration statements — has moved from a compliance buzzword to an active enforcement priority. The SEC does not have a specific AI-washing statute, and it does not need one. The agency is using existing anti-fraud authority that has been in place for decades, and it is applying that authority to AI-related misstatements the same way it has applied it to every other category of material misrepresentation.

For public companies, startup founders preparing registration statements, and any business that describes AI capabilities to investors, this post explains how the SEC’s enforcement theory works, what the recent cases establish, and what a legally defensible AI disclosure program looks like.

The Legal Framework: No New Rules Required

The SEC’s ability to pursue AI-washing cases rests entirely on existing authority. Understanding why requires a brief review of the anti-fraud provisions that apply to public company disclosures.

Rule 10b-5

Rule 10b-5, promulgated under Section 10(b) of the Securities Exchange Act of 1934, prohibits any person from making a material misstatement or omission in connection with the purchase or sale of any security. The elements of a 10b-5 violation include: (1) a material misstatement or omission; (2) scienter (for SEC enforcement, negligence or recklessness is sufficient in some contexts); (3) in connection with the purchase or sale of a security; and (4) reliance (established presumptively under the “fraud on the market” doctrine for publicly traded securities).

A statement about a company’s AI capabilities is material — and therefore potentially actionable — when there is a substantial likelihood that a reasonable investor would consider it important in making an investment decision. In an environment where AI capabilities are among the most significant drivers of technology company valuations, that bar is not high. Investors are paying close attention to AI capability claims, and the SEC knows it.

Section 17(a) of the Securities Act

Section 17(a) applies in the context of the offer or sale of securities. It has three subsections: Section 17(a)(1) requires scienter; Section 17(a)(2) and (3) can be established through negligence. This means that for registration statements and securities offerings, the SEC can pursue AI-washing charges without proving intentional fraud — negligent misrepresentation about AI capabilities is enough.

The Presto Automation enforcement action in January 2025 relied in part on Section 17(a)(2), which prohibits obtaining money or property by means of a material misstatement or omission. This is the lower-scienter provision, and its use in that case signals that the SEC is not limiting its AI-washing theory to cases of deliberate fraud.

Marketing Rule (Rule 206(4)-1)

For investment advisers, the Marketing Rule under the Investment Advisers Act prohibits advertisements that include materially misleading statements. The March 2024 enforcement actions against Delphia and Global Predictions both charged Marketing Rule violations alongside fraud charges. This rule reaches marketing materials, pitch decks, and investor presentations — the same channels through which AI claims are most commonly made.

Compliance Rule (Rule 206(4)-7)

The Compliance Rule requires investment advisers to adopt and implement written policies and procedures reasonably designed to prevent violations of the Advisers Act. Enforcement actions against advisers for AI washing have also included Compliance Rule violations on the theory that a compliant compliance program would have caught and corrected the misleading AI claims before they were disseminated.

The Enforcement Cases: What They Establish

March 2024: Delphia (USA) Inc. — $225,000 Penalty

The SEC’s first explicit AI-washing enforcement actions, announced on March 18, 2024, targeted two investment advisers. The Delphia case is the more widely discussed.

Delphia marketed itself as an AI-driven investment adviser that used client personal data — app usage, transaction records, and similar data — to power investment decisions. In a 2019 press release, Delphia claimed it was “the first investment adviser to convert personal data into a renewable source of investable capital.”

The problem: Delphia had not built the algorithm it claimed to have built. The firm had not used client data in the manner described. The AI-powered investment process was a marketing claim without operational substance behind it.

The SEC charged Delphia with violations of Section 206(2) and 206(4) of the Investment Advisers Act, as well as the Marketing Rule and the Compliance Rule. Delphia paid a $225,000 civil penalty.

What this case establishes: A company does not need to be a public company, and the misstatement does not need to appear in an SEC filing. Marketing materials — press releases, investor presentations, website claims — are sufficient to trigger liability if they contain material misstatements about AI capabilities.

March 2024: Global Predictions, Inc. — $175,000 Penalty

The companion enforcement action to Delphia targeted Global Predictions, another registered investment adviser. Global Predictions made claims about its “AI-driven” forecasting and represented that it used specific AI methodologies in ways that were inaccurate or could not be substantiated.

The SEC charged Global Predictions with negligent fraud under Section 206(2) and the Marketing Rule. The $175,000 penalty, while modest, was notable as part of the SEC’s first coordinated AI-washing enforcement announcement.

What this case establishes: The SEC was willing to bring enforcement actions based on the negligent fraud standard — recklessness about the accuracy of AI claims is sufficient. A firm does not need to have deliberately fabricated its AI capabilities; failing to ensure that AI-related marketing claims are accurate and substantiated is enough.

January 2025: Presto Automation Inc. — Public Company First

The Presto Automation case, settled in January 2025, marked the first SEC AI-washing enforcement action against a public company. Presto was a formerly Nasdaq-listed restaurant technology company that marketed its “Presto Voice” product as an AI-powered drive-through ordering system.

The SEC’s charges included:

• From November 2021 to September 2022, the speech recognition technology in all commercially deployed units of Presto Voice was actually owned and operated by a third-party supplier — not by Presto — and this was not disclosed to investors.
• When Presto eventually deployed its own proprietary technology, it told investors the system “eliminated the need for human order taking.” In reality, the system could not take orders independently and required substantial human involvement.
• Presto employed off-site human agents, including workers in the Philippines and India, to process orders that the AI could not handle — a fact hidden from investors.

Presto consented to a cease-and-desist order. The case was settled without admission or denial, but the factual findings are detailed and damning.

What this case establishes: Public company disclosure failures around AI — in earnings materials, investor presentations, and SEC filings — are subject to the same anti-fraud enforcement as any other material misstatement. The case also establishes that concealing the human labor behind an apparently AI-powered product is an independent disclosure violation, not merely a technology marketing issue.

April 2025: Nate Inc. — Criminal Charges

The Nate Inc. case crossed from civil SEC enforcement into criminal charges. The SEC and the U.S. Attorney’s Office for the Southern District of New York filed parallel actions against Albert Saniger, the founder and former CEO of Nate, alleging he raised over $42 million by falsely claiming Nate’s shopping app processed transactions through AI, machine learning, and neural networks. In reality, the transactions were processed manually by contract workers in foreign countries. Saniger allegedly fabricated automation rate metrics, claiming rates above 90% when the actual rate was essentially zero.

What this case establishes: When AI-washing crosses into deliberate fabrication — inventing performance metrics, hiding the true nature of the technology from investors — it is securities fraud, and it carries criminal exposure. The DOJ involvement signals that prosecutors are treating sophisticated AI-related investment fraud as a serious criminal matter.

What “AI Washing” Looks Like in Practice

AI-washing enforcement does not require outright fabrication. The cases establish that the following practices can trigger liability:

Attributing capabilities to AI that the company does not possess. Claiming that AI drives investment decisions, processes transactions, or automates workflows when those processes are manual or rely on third-party technology is a misstatement.

Using AI terminology without substantiation. Describing a product as “AI-powered,” “machine learning-driven,” or “AI-enabled” when the technology does not have those features, or when the AI component plays a marginal role in the described function, risks being characterized as materially misleading.

Omitting third-party dependency. If a company’s AI capability is provided entirely by a third-party vendor and the company presents it as proprietary, the failure to disclose the third-party dependency may be a material omission.

Exaggerating AI performance metrics. Claiming specific accuracy rates, automation rates, or performance statistics that are not supported by actual system data is a misstatement.

Disclosing AI risks while overstating AI capabilities. Some companies are attempting to manage both sides of the AI narrative — listing AI as a risk factor in their 10-K while simultaneously claiming robust AI capabilities in investor presentations. The inconsistency between filings and marketing materials is itself a red flag that examination staff have flagged.

The 2026 Disclosure Environment

Seventy-six percent of S&P 500 companies added or expanded AI as a material risk description in their 2025 annual filings. Among Fortune 500 companies, approximately 84% discussed AI in recent Form 10-Ks. The volume of AI disclosure has increased dramatically — but quality and accuracy have not kept pace.

The SEC’s examination staff have signaled several specific areas of scrutiny for AI-related disclosures:

Specificity versus boilerplate. Generic risk factors that list AI as a risk without explaining how AI is actually used in the company’s business are being questioned. Examination staff want to understand what AI actually does in the company’s operations, not just that AI exists as a category of risk.

Consistency between filings and other channels. The SEC cross-references public company SEC filings with earnings call transcripts, investor day presentations, and marketing materials. Inconsistencies between the measured language in formal filings and the expansive claims in investor presentations are exactly what examination staff look for.

Substantiation for affirmative AI claims. When a company affirmatively claims AI capabilities — in the business description, the MD&A, or earnings guidance — examination staff may ask what documentation supports those claims.

Building a Legally Defensible AI Disclosure Program

Public companies and companies preparing for capital raises should build AI disclosure programs around the following principles:

Accuracy review for all AI claims. Every statement about AI capabilities made in a public forum — SEC filings, earnings calls, investor presentations, press releases, website copy — should be reviewed by someone with technical knowledge of the actual system. Marketing language and technical reality need to be reconciled before publication.

Internal documentation of AI capabilities. For any AI system described to investors, the company should maintain internal documentation of what the system actually does, what data it processes, what decisions it influences, what human oversight it requires, and what its actual performance metrics are. This documentation is the evidentiary foundation if an enforcement inquiry is ever opened.

Disclosure of third-party dependency. If AI capabilities are provided by a third-party vendor, that dependency should be disclosed. The Presto case makes clear that presenting third-party technology as proprietary is a disclosure failure.

Consistency review across channels. All AI-related public statements — from informal social media posts by executives to formal SEC filings — should be reviewed for consistency. The SEC does not limit its analysis to formal filings.

Conservative language calibration. Legal and compliance review of AI-related marketing claims should apply the “would a reasonable investor consider this important?” standard. If a claim would influence an investor’s decision, it needs to be accurate and substantiated.

Board-level oversight. For public companies, board oversight of AI risk and AI-related disclosures is increasingly expected. Audit committees and risk committees should have visibility into AI capability claims made to investors and the documentation supporting those claims.

Separate the hype from the filing. Startup founders and executives often communicate about AI capabilities in terms calibrated for fundraising enthusiasm rather than legal precision. The SEC’s cases demonstrate that those communications — even when not in SEC filings — are fair game for enforcement. Internal communications discipline around AI claims is a real compliance obligation, not just good practice.

The Trend Line

The SEC’s AI-washing enforcement record from 2024 through 2025 — five enforcement actions, criminal charges in at least one case, and clear signals that more are coming — establishes an enforcement trajectory that is moving in one direction. The cases to date have been the relatively straightforward ones: companies that claimed AI capabilities they demonstrably did not have. The harder cases — companies whose AI claims were technically accurate but materially misleading — are likely in the pipeline.

Companies that take a rigorous approach to AI disclosure now, grounding every public claim in documented technical reality, will be positioned to defend their disclosures if the SEC comes knocking.

This post is for general informational purposes only and does not constitute legal advice. Reading this post does not create an attorney-client relationship. If you have questions about your specific situation, consult a qualified attorney.