Insights Article
Get the latest insights and updates delivered to your inbox.
Last quarter, S&P 500 companies that cited "AI" on their earnings calls outperformed those that did not by as much as 8.2 percentage points.¹ This quarter, the relationship reversed.
According to FactSet data published on March 12, 2026, S&P 500 companies that cited "AI" on their fourth-quarter 2025 earnings calls returned an average of 1.5 percent from December 31, 2025, through March 10, 2026. Companies that did not cite AI returned 5.6 percent.² In a single quarter, the market flipped from rewarding AI rhetoric to punishing it.
Something has changed. The question institutional investors must now confront is what—and whether they have been paying close enough attention to see it coming.
The simplest explanation is that the market is beginning to distinguish between companies that are genuinely building artificial intelligence capabilities and companies that are merely saying the words. That distinction has a name. Regulators call it "AI washing"—the practice of making materially false, misleading, or exaggerated claims about a company's AI capabilities in public filings, earnings calls, investor presentations, and marketing materials. And it has become, by any measure, one of the most consequential disclosure risks facing public companies and the investors who own them.
The numbers suggest that the AI disclosure environment has become saturated to the point of meaninglessness—and that the saturation is creating fertile conditions for misrepresentation.
FactSet's latest data shows that a record 331 S&P 500 companies cited "AI" on their Q4 2025 earnings calls—68 percent of all calls conducted during the period, and the highest figure in at least a decade.³ That represents a six-fold increase from the quarter before OpenAI launched ChatGPT in November 2022.⁴ In the Information Technology and Communication Services sectors, the penetration rate has reached 94 and 89 percent, respectively.³ Even the Financials sector, where artificial intelligence has historically been discussed in narrow operational contexts, saw 91 percent of its earnings calls reference AI in the fourth quarter.³
When two-thirds of the S&P 500 is claiming AI relevance in every earnings call, the signal-to-noise ratio collapses. For institutional investors conducting fundamental analysis, the critical challenge is no longer identifying which companies are pursuing AI. It is identifying which ones are telling the truth about it.
The enforcement record suggests that a meaningful number are not.
Get the latest insights delivered to your inbox.
The Securities and Exchange Commission's first AI washing enforcement actions were brought in March 2024, when the agency settled charges against two registered investment advisers—Delphia Inc. and Global Predictions Inc.—for making false and misleading claims about their use of artificial intelligence.⁵ Delphia had told investors in SEC filings and marketing materials that it used AI and machine learning to incorporate client data into its investment process. Global Predictions had claimed to be the "first regulated AI financial advisor" and described its platform as providing AI-driven forecasts. The SEC found that both firms' AI claims were materially overstated.⁵ The cases were modest in scale—Delphia paid a $225,000 civil penalty and Global Predictions paid $175,000—but the Commission's message was unmistakable. As then-Enforcement Director Gurbir Grewal stated: "If you claim to use AI in your investment processes, you need to ensure that your representations are not false or misleading."⁵
The enforcement actions that followed were neither modest nor subtle.
In January 2025, the SEC brought a settled enforcement action against Presto Automation Inc., a restaurant-technology company that had been listed on the Nasdaq, for making materially false and misleading statements about its flagship product, Presto Voice.⁶ The company had marketed Presto Voice as an AI-powered speech recognition system capable of automating drive-through order-taking at quick-service restaurants. In SEC filings between 2021 and 2023, Presto represented the technology as though it were internally developed and stated that it "eliminat[ed] human order taking."⁶ The SEC found that the technology was owned and operated by a third party, and that Presto "relied heavily" on human employees located in the Philippines and India to complete orders.⁷ Over 70 percent of orders processed through the in-house version of Presto Voice required human intervention, and at certain locations, 100 percent of orders required human intervention.⁷ The company had entered the public markets through a 2022 SPAC merger—a detail that underscores the elevated disclosure risks associated with the de-SPAC pathway, where companies with limited public-market operating histories make forward-looking claims to investors that may not survive scrutiny.
In April 2025, the SEC and the U.S. Attorney's Office for the Southern District of New York filed parallel civil and criminal actions against Albert Saniger, the founder and former CEO of Nate Inc., a privately held technology startup.⁸ Saniger had raised over $42 million from venture capital investors by marketing Nate as a mobile shopping application that used proprietary AI, machine learning, and neural networks to complete online purchases with a single tap.⁸ He described the technology as working "like magic."⁹ The DOJ alleged that the actual automation rate was "effectively zero percent."⁹ Purchases were not completed by artificial intelligence. They were completed by hundreds of contract workers in call centers in the Philippines and Romania, manually processing transactions that users and investors believed were automated.⁹ The SEC alleged that Saniger personally profited from the fraud, including by selling approximately $3 million of his own shares during a June 2021 fundraising round.¹⁰ Nate ceased operations in January 2023, and investors lost tens of millions of dollars.¹⁰
These cases are notable not only for their facts—which border on the farcical—but for what they represent structurally. The Nate Inc. prosecution marked the first AI washing enforcement action brought by both the SEC and the DOJ under the current administration, signaling bipartisan continuity in the treatment of AI-related fraud as a priority regardless of which party controls the Commission.¹¹
The enforcement actions are accompanied by a parallel surge in private securities litigation that suggests the plaintiffs' bar has identified AI washing as a durable and expanding source of claims.
According to data from the Stanford Law School Securities Class Action Clearinghouse, more than 53 securities class action lawsuits involving AI-related allegations have been filed since the first such case in March 2020.¹² The trajectory is accelerating. Seven AI-related securities class actions were filed in 2023. That number doubled to 14 in 2024. By the first half of 2025 alone, 12 additional cases had been filed.¹³
DLA Piper, in a September 2025 analysis, identified a recurring pattern across these complaints: companies allegedly overstated the sophistication, effectiveness, or uniqueness of their AI capabilities, and the litigation followed when operational results failed to match the claims.¹³ Several of the cases were initiated after short-seller reports—a pattern that connects directly to the detection gap we identified in our prior Insights article on the disappearance of activist short sellers.¹⁴
The dollar exposure across the broader securities class action landscape—within which AI-related cases are a growing component—underscores the stakes. Cornerstone Research reported that the Disclosure Dollar Loss Index for all securities class actions—measuring aggregate investor losses based on stock price declines following corrective disclosures—reached $403 billion in the first half of 2025, a 56 percent increase from the second half of 2024.¹⁵ The Maximum Dollar Loss Index surged to $1.851 trillion, a 154 percent increase, driven in significant part by "mega-litigation" targeting large-capitalization companies.¹⁵ Stanford Law Professor and former SEC Commissioner Joseph Grundfest, commenting on the intersection of these trends with AI-related claims, characterized the landscape succinctly: "The big news in these trends are the dollars at risk and AI."¹⁵
The SEC has not been content to pursue AI washing through case-by-case enforcement alone. It has built institutional infrastructure specifically designed to sustain and escalate scrutiny.
In 2025, the Commission rebranded its Crypto Assets and Cyber Unit as the Cyber and Emerging Technologies Unit, or CETU—a reorganization that expanded the unit's mandate beyond cryptocurrency to encompass AI-related misconduct, cyber intrusions, and social-media-driven fraud.¹⁶ Senior officials from CETU and the Division of Enforcement stated publicly at the Securities Enforcement Forum West in May 2025 that "rooting out" AI washing fraud schemes was an "immediate priority."¹⁷
In November 2025, the SEC's Division of Examinations released its fiscal year 2026 examination priorities, which explicitly identified the assessment of registrants' AI-related representations for accuracy and the adequacy of firms' policies and procedures governing the use of AI as focus areas.¹⁸ And on February 24, 2026, the SEC published its first comprehensive update to the Enforcement Manual since 2017—a revision that, while not AI-specific, formalizes procedural reforms that will facilitate more efficient investigation and resolution of disclosure fraud cases broadly, including those involving AI-related misrepresentations.¹⁹
For institutional investors, the regulatory trajectory is clear. This is not an enforcement priority that will recede. The SEC has built a dedicated unit, published examination priorities, and updated its procedural infrastructure. The DOJ has demonstrated willingness to pursue parallel criminal charges. And private plaintiffs have identified AI washing as a litigation category with significant potential recoveries. Companies making AI claims that cannot withstand forensic scrutiny are operating in an environment of escalating legal exposure on multiple fronts simultaneously.
At Buxton Helmsley, our forensic approach to financial statement analysis has always emphasized that the most consequential misrepresentations are found not in what companies say, but in the gap between what they say and what the data shows. AI washing is a textbook application of this principle. The claims are public. The data to evaluate them is, in most cases, available to any investor willing to look.
The first category of forensic indicators involves the relationship between AI claims and research and development expenditure. A company that describes AI as central to its product strategy, growth trajectory, or competitive positioning should be spending commensurately on research and development. When a company's AI narrative becomes more aggressive on earnings calls while R&D expenditure as a percentage of revenue remains flat or declines, the discrepancy warrants investigation. The forensic investor should compare R&D spending trends against the intensity and specificity of management's AI claims over the same period. A widening gap between rhetoric and investment is a leading indicator.
The second category involves revenue attribution. When management claims that AI is driving revenue growth, the investor should test that claim against observable revenue composition. Is the company disclosing AI-related revenue separately? If not, why not—and what does that tell us about the materiality of the AI contribution? Companies that describe AI as transformative but cannot isolate its contribution to revenue may be conflating correlation with causation, or worse, attributing revenue to AI capabilities that do not yet exist at commercial scale.
The third category involves workforce composition and capital expenditure patterns. Artificial intelligence at scale requires specific infrastructure investments: GPU clusters, cloud computing capacity, data engineering talent, and machine learning engineers. A company claiming to deploy AI capabilities should show corresponding evidence in its capital expenditure disclosures, cloud computing commitments, and workforce composition. When a company's AI narrative is aggressive but its capital expenditure profile is indistinguishable from a pre-AI peer, the forensic investor should ask what infrastructure is supporting the claimed capabilities. The Presto Automation and Nate Inc. cases both illustrate this pattern—companies that claimed AI automation while relying on human labor that would have been visible in operating cost structures to any analyst who looked.
The fourth category involves the specificity of disclosure. Companies with genuine AI capabilities tend to disclose specific technical details: which models they are deploying, what training data they are using, how they are measuring performance, and what limitations the technology has. Companies engaged in AI washing tend to use vague, aspirational language—"AI-powered," "leveraging machine learning," "integrating artificial intelligence"—without providing technical specifics that would allow an investor to evaluate the claim independently. The SEC's orders in both the Delphia and Presto cases identified precisely this pattern: broad, unqualified claims unsupported by operational reality.
The fifth category involves auditor scrutiny of AI-related asset valuations and intangible asset recognition. As companies capitalize software development costs under ASC 350-40 or recognize internally developed AI-related intangible assets, the valuation assumptions underlying those capitalizations should be tested against the company's actual deployment of the technology. A company that is capitalizing significant AI development costs while the underlying technology has not reached commercial viability—or, worse, does not function as described—is potentially misstating both its asset base and its operating expenses.
The AI washing phenomenon is not, at its core, a technology problem. It is a disclosure problem—and one that shares structural DNA with every major category of securities fraud that has preceded it.
In the 1990s, companies exaggerated their internet capabilities during the dot-com bubble, creating valuations that had no relationship to actual revenue or operating performance. In the 2000s, companies misrepresented the risk profiles of structured financial products, contributing to a systemic crisis. In the 2010s, companies engaged in greenwashing—exaggerating environmental commitments to attract ESG-oriented capital. In each cycle, the pattern was the same: a powerful narrative attracted investor capital, companies exaggerated their participation in that narrative to capture more capital, and the gap between claims and reality eventually produced enforcement actions, litigation, and investor losses.
AI washing follows this pattern precisely. The difference is that the narrative is more powerful than any that preceded it, the capital flows are larger, and the gap between what companies claim and what they can actually do is, in many cases, wider. When 68 percent of the S&P 500 is claiming AI relevance on its earnings calls, and regulators are simultaneously prosecuting companies for fabricating AI capabilities out of whole cloth, the market is sending a message that institutional investors cannot afford to ignore.
The companies that are genuinely building AI capabilities will be differentiated by the specificity of their disclosures, the consistency between their claims and their capital expenditure patterns, and their willingness to discuss limitations alongside aspirations. The companies that are merely washing their disclosures with AI language will be differentiated by enforcement actions, securities litigation, and the slow erosion of investor confidence that occurs when the gap between narrative and reality becomes too large to sustain.
The forensic investor's task is to identify which category a company falls into before the market—or the SEC—makes the determination for them.
The record 331 S&P 500 companies citing AI on their most recent earnings calls are not all lying. Many of them are making genuine investments in transformative technology that will reshape their industries over the coming decade. But the enforcement record, the litigation trajectory, and the sheer saturation of AI-related claims in corporate disclosures make one thing clear: the signal has been diluted to the point where uncritical acceptance of AI narratives is no longer a defensible investment practice.
The SEC has built the institutional apparatus to prosecute AI washing. The plaintiffs' bar has identified it as a litigation category with substantial recoveries. And the market itself is beginning to punish companies for mentioning AI without demonstrating substance behind the words.
For institutional investors, the implication is straightforward. Treat every AI claim in a corporate filing, earnings call, or investor presentation with the same forensic skepticism that would be applied to any other material representation about the business. Test the claim against the data. Compare the narrative against capital expenditure. Evaluate whether the company can isolate AI's contribution to revenue. And remember that the four most expensive words in a prospectus have always been "this time is different"—whether the subject was the internet in 1999, structured products in 2007, or artificial intelligence in 2026.
The companies that are genuinely powered by AI have nothing to fear from scrutiny. The ones that are powered by artificial claims will not survive it.
Get the latest insights delivered to your inbox.
Referenced Sources
[1] FactSet, "Highest Number of S&P 500 Earnings Calls Citing 'AI' Over the Past 10 Years" (December 8, 2025), reporting that S&P 500 companies citing "AI" on Q3 2025 earnings calls outperformed non-citers, with average returns of 13.9 percent versus 5.7 percent since December 31, 8.1 percent versus 3.9 percent since June 30, and 1.0 percent versus 0.3 percent since September 30.
[2] FactSet, "More Than 65% of S&P 500 Earnings Calls for Q4 Cited 'AI'" (March 12, 2026), reporting that S&P 500 companies citing "AI" on Q4 2025 earnings calls saw an average price increase of 1.5 percent from December 31, 2025, through March 10, 2026, compared to 5.6 percent for companies that did not cite "AI."
[3] FactSet (March 12, 2026), supra note 2. A record 331 S&P 500 companies cited "AI" on Q4 2025 earnings calls, representing 68 percent of the 485 earnings calls conducted during the period. The Information Technology sector reached 94 percent, Communication Services 89 percent, and Financials 91 percent.
[4] Based on FactSet's historical tracking of AI mentions on S&P 500 earnings calls, as reported in Inc., "S&P 500 mentions of 'AI' on earnings calls just hit a record high" (December 12, 2025), which noted that the Q3 2025 figure of 306 represented "six times the amount compared to the quarter before ChatGPT launched in November 2022." The Q4 2025 figure of 331, reported by FactSet on March 12, 2026, represents an even larger multiple.
[5] U.S. Securities and Exchange Commission, "SEC Charges Two Investment Advisers with Making False and Misleading Statements About Their Use of Artificial Intelligence," Press Release No. 2024-36 (March 18, 2024). Delphia Inc. agreed to pay a $225,000 civil penalty; Global Predictions Inc. agreed to pay a $175,000 civil penalty.
[6] U.S. Securities and Exchange Commission, "SEC Charges Restaurant-Technology Company Presto Automation for Misleading Statements About AI Product," Release No. 33-11352 (January 14, 2025).
[7] Lowenstein Sandler LLP, "SEC Charges Public Company with AI Washing" (February 3, 2025), reporting that over 70 percent of orders processed through the in-house version of Presto Voice required human intervention, and 100 percent of orders at certain locations required human intervention.
[8] U.S. Securities and Exchange Commission, "Alberto Saniger Mantinan, a/k/a Albert Saniger," Litigation Release No. LR-26282 (April 9, 2025); U.S. Attorney's Office, Southern District of New York, Indictment, United States v. Albert Saniger (unsealed April 11, 2025).
[9] Fortune, "A tech CEO has been charged with fraud for saying his e-commerce startup was powered by AI, when it was actually just using manual human labor" (April 11, 2025), reporting that the DOJ alleged the actual automation rate was "effectively zero percent" and that Saniger hired hundreds of overseas contractors to complete purchases.
[10] The D&O Diary, "Tech Exec Charged with AI Washing-Related Securities Fraud" (April 13, 2025), citing the SEC's allegation that Saniger personally sold approximately $3 million of Nate shares during a June 2021 fundraising round and that investors were left with tens of millions of dollars in losses when Nate dissolved in January 2023.
[11] Debevoise & Plimpton LLP, "The SEC and DOJ Signal Continued Focus on AI Washing Under Trump Administration" (April 17, 2025), noting that the Nate Inc. enforcement actions were "the first AI-washing enforcement actions brought by the SEC and Department of Justice under the new Trump administration."
[12] WTW, "More buzz than sting: The state of AI-related securities litigation" (November 14, 2025), citing the Stanford Law School Securities Class Action Clearinghouse, which identified 53 securities class actions related to AI through H1 2025.
[13] DLA Piper, "AI-related securities class action filings are on the rise: Key observations" (September 2025), reporting that AI-related securities class actions doubled from 7 in 2023 to 14 in 2024, with 12 additional cases filed through the first half of 2025, and identifying a recurring pattern of companies allegedly overstating the sophistication, effectiveness, or uniqueness of their AI capabilities.
[14] See Buxton Helmsley, "The Last Line of Defense: How the Disappearance of Activist Short Sellers Is Leaving Corporate Fraud Undetected—and What Institutional Investors Must Do About It," Insights (March 2026).
[15] Cornerstone Research, "Securities Class Action Filings—2025 Mid-Year Assessment" (July 29, 2025), reporting that the Disclosure Dollar Loss Index reached $403 billion in H1 2025 (56 percent increase from H2 2024) and the Maximum Dollar Loss Index surged to $1.851 trillion (154 percent increase). Stanford Law Professor and former SEC Commissioner Joseph Grundfest quoted: "The big news in these trends are the dollars at risk and AI."
[16] Cleary Gottlieb, "The Shifting SEC Enforcement Landscape: 2025 Year-in-Review" (January 2026), noting that the SEC rebranded its Crypto Assets and Cyber Unit as the Cyber and Emerging Technologies Unit (CETU), with a mission of "combatting cyber-related misconduct and to protect retail investors from bad actors in the emerging technologies space."
[17] DLA Piper, "SEC emphasizes focus on 'AI washing' despite perceived enforcement slowdown" (2025), reporting that senior officials from the SEC's Enforcement Division and CETU reiterated at the Securities Enforcement Forum West on May 15, 2025, that "rooting out" AI washing fraud schemes was an "immediate priority."
[18] U.S. Securities and Exchange Commission, Division of Examinations, "2026 Examination Priorities" (November 17, 2025), identifying the assessment of registrants' AI-related representations and the adequacy of firms' AI-related policies and procedures as focus areas.
[19] Cooley LLP, "Updated SEC Enforcement Manual Emphasizes Engagement and Transparency" (February 27, 2026), noting that the February 24, 2026, update was the first comprehensive revision since 2017.
This article is published by Buxton Helmsley USA, Inc. for informational and educational purposes only. It does not constitute investment advice, a solicitation, or a recommendation to buy or sell any security. The views expressed are those of Buxton Helmsley and are based on publicly available information as of the date of publication. Investors should conduct their own due diligence and consult with qualified professionals before making investment decisions.
Insights and updates delivered directly to your inbox.