The SEC’s AI washing enforcement theory, first used against advisers, now targets founders whose pitch decks overstate AI automation to investors.
AI Washing Enforcement: Securities Fraud Exposure for Pitch Decks and Investor Updates
Executive Summary
Overstating AI automation to investors exposes founders to securities fraud liability under Rule 10b-5, the theory the SEC used to fine Delphia and Global Predictions four hundred thousand US Dollars. Founders who describe a product as AI powered when it relies mainly on human review, who overstate how far an internal model has progressed, or who allow a staff member to complete an order by hand while an investor watches an automated demonstration, create the same exposure the Securities and Exchange Commission applied to two registered investment advisers in March 2024. The Commission has since applied the identical theory to operating companies raising private capital. It charged Nate Inc. founder Albert Saniger in April 2025 for describing manually processed purchases as autonomous artificial intelligence, and it charged Joonko Diversity Inc. founder Illit Raz in June 2024 for fabricating the sophistication of a recruiting algorithm. Both matters carry parallel criminal charges from the Department of Justice.
The legal theory is settled, even where the specific wording that satisfies it is not. Securities Act Section 17(a) and Exchange Act Section 10(b) and Rule 10b-5 prohibit a material misstatement made in connection with the offer or sale of any security, a category that includes a Regulation D private placement pitched from a slide deck as much as a registered public offering. What remains genuinely open is the line between an aspirational product roadmap, which the law permits, and a present tense factual claim about current capability, which the law treats as a representation an investor is entitled to rely on. The Commission formed a Cyber and Emerging Technologies Unit in February 2025 specifically to pursue this category of case, and has named artificial intelligence misrepresentation an immediate enforcement priority heading into the 2026 reporting season.
Detailed Findings
What securities law governs an AI claim in a pitch deck
A statement in a pitch deck or investor update answers to the same anti fraud statutes that govern a public company’s disclosures, because Securities Act Section 17(a), 15 U.S.C. Section 77q(a), and Exchange Act Section 10(b) and Rule 10b-5, 15 U.S.C. Section 78j(b) and 17 C.F.R. Section 240.10b-5, reach fraud in the offer or sale of any security and fraud in connection with the purchase or sale of any security. Neither provision carves out securities sold privately under Regulation D. A Regulation D exemption removes the obligation to register an offering with the Commission. It does not remove the obligation to tell the truth about the product being sold.
Both Saniger and Raz built companies that never registered a public offering, and the Commission charged each of them under Section 17(a) and Section 10(b) and Rule 10b-5, confirming that a Regulation D raise carries full anti fraud exposure. A slide deck used to raise a seed or Series A round is a written communication made in connection with the sale of a security, and every material sentence on it becomes a statement of fact capable of supporting a fraud claim if it turns out to be false. In reviewing investor materials for financings this year, the recurring gap is rarely an outright lie. It is an unqualified verb. A slide states that a product uses artificial intelligence without saying what share of the output a person reviews or edits before a customer sees it, and when the founding team is asked for that percentage directly, most cannot supply one.
The founding AI washing cases: Delphia and Global Predictions
In March 2024 the Commission used the Investment Advisers Act Marketing Rule to fine two registered investment advisers a combined four hundred thousand US Dollars for describing artificial intelligence capability neither firm actually had, establishing the enforcement theory the Commission has since carried into cases against operating companies. Delphia (USA) Inc. told the Commission and told investors between 2019 and 2023 that it put collective client data to work to make its artificial intelligence smarter, when the firm had not developed the machine learning capability it described. The Commission fined Delphia two hundred twenty five thousand US Dollars.
Global Predictions Inc. marketed itself in 2023 as the first regulated artificial intelligence financial advisor and promised expert forecasts it described as driven by artificial intelligence, along with a tax loss harvesting service it did not actually provide. The Commission fined Global Predictions one hundred seventy five thousand US Dollars. Both orders rested on Sections 206(2) and 206(4) of the Investment Advisers Act of 1940 and the Marketing Rule, Rule 206(4)-1, which governs advertisements by registered investment advisers. The Commission’s chair at the time stated that describing an artificial intelligence capability a firm does not have hurts investors in the same way any other false marketing claim does. The order framed the violation as specific to registered advisers, but the underlying theory, that a specific and checkable technical claim made to attract investor money must be true, carried directly into cases against companies with no investment adviser registration at all.
Extension to operating companies: Presto Automation and negligence based liability
The Commission does not need to prove intent to defraud to charge an operating company with AI washing, because Presto Automation Inc. settled AI washing charges in January 2025 under a provision that required no showing that anyone at the company meant to deceive investors. Presto, a publicly traded restaurant technology company, told investors its voice ordering product used the company’s own speech recognition technology and had reached an automated order completion rate as high as 95 percent, eliminating the need for a person to take an order. Until September 2022 the product ran entirely on a supplier’s technology rather than Presto’s own, and the version Presto did build still required a human agent to step in for roughly 70 percent of orders through December 2023. The reported 95 percent figure excluded intervention by anyone other than restaurant staff, which understated how much of the order handling remained manual.
The Commission charged a violation of Securities Act Section 17(a)(2), a provision that prohibits obtaining money by means of a material misstatement and does not require proof of intent, along with Exchange Act Section 13(a) and Rules 13a-1 and 13a-15(a), which govern periodic reporting and disclosure controls. The Commission imposed a cease and desist order with no civil penalty, citing Presto’s financial condition and its remedial steps, including new disclosure controls built specifically to check artificial intelligence claims before they reach investors. A company does not need to have lied on purpose to face an enforcement action. Failing to build a process that checks a technical claim against what the product can actually do is itself a violation.
Founder level fraud and criminal exposure: Nate Inc. and Joonko Diversity
Two matters charged in 2024 and 2025 move the AI washing theory from administrative settlement to parallel criminal prosecution, exposing a founder personally to a federal prison sentence in addition to Commission penalties. Nate Inc. founder Albert Saniger raised approximately 42 million US Dollars between 2019 and December 2022 by telling investors that the Nate shopping application completed purchases automatically using artificial intelligence, when contractors in the Philippines and Romania processed the transactions by hand, later replaced by less capable scripted tools. The Commission’s civil complaint, filed April 9, 2025, alleges Saniger directed employees to process orders manually behind the scenes during investor demonstrations to create the appearance of automation. The Department of Justice filed parallel criminal charges the same day in the Southern District of New York for securities fraud and wire fraud, each carrying a statutory maximum sentence of 20 years.
Joonko Diversity Inc. founder Illit Raz raised 21 million US Dollars in 2021 and 2022 by describing a recruiting product built on seven different proprietary algorithms said to use natural language processing, computer vision, and machine learning to match candidates to employers automatically, and by reporting more than 200 corporate customers, 185,000 active monthly candidates, and revenue climbing toward eight and a half million US Dollars by mid 2023. The Commission and the United States Attorney’s Office for the Southern District of New York found these figures fabricated when they charged Raz on June 11, 2024. Both matters rest on Securities Act Section 17(a) and Exchange Act Section 10(b) and Rule 10b-5, the general anti fraud provisions that apply regardless of whether the issuer is publicly traded. Staging a manual process to look automated during a live demonstration, rather than merely describing the technology inaccurately in writing, reads as affirmative fraud rather than a disclosure gap, and it is the fact pattern most likely to draw a parallel criminal referral instead of a civil settlement alone.
Who is personally liable: officers, statement makers, and control persons
An officer who signs, drafts, or personally delivers an AI claim to investors can be individually charged under the same anti fraud provisions as the company, and an executive who controls the company can face additional liability even without directly making the misstatement. The Supreme Court held in Janus Capital Group, Inc. v. First Derivative Traders, 564 U.S. 135 (2011), that primary liability under Rule 10b-5 for making a false statement attaches to the person with ultimate authority over the statement, including its content and whether and how to communicate it, a standard that reaches a founder or chief executive who personally builds and presents a pitch deck.
Section 20(a) of the Exchange Act, 15 U.S.C. Section 78t(a), separately extends liability to any person who controls a company found liable for securities fraud, subject to a good faith defense. A board member or chief technology officer who reviewed and approved an inflated claim without speaking it directly can still face exposure under this provision. In both Nate Inc. and Joonko Diversity, the Commission and the Department of Justice charged the founder personally rather than only the corporate entity, a pattern that should inform who signs off on investor facing artificial intelligence language and how that approval gets documented inside the company.
The greenwashing precedent the Commission is following
The Commission is applying to artificial intelligence claims the same enforcement model it built between 2022 and 2023 for environmental, social, and corporate responsibility claims, so that earlier enforcement record is the closest available guide to how AI washing cases will be charged and settled going forward. In May 2022 the Commission fined BNY Mellon Investment Adviser one and a half million US Dollars for representing that all investments in certain funds had undergone a quality review tied to those factors when many had not, a misstatement about an internal screening process rather than about investment performance. The order cited Sections 206(2) and 206(4) of the Investment Advisers Act.
In November 2022 the Commission fined Goldman Sachs Asset Management four million US Dollars for failing to follow its own written investment policies and procedures for screening those same factors, an enforcement theory built on the gap between a public description of a process and the internal record of what the process actually was. The environmental and social claims cases and the AI washing cases share a structural feature: a specific, checkable description of an internal process, a screening step, or a technical capability that a regulator can compare directly against internal records and product logs. The same gap between polished external language and an untested internal reality supports either category of case.
The enforcement posture entering the 2026 reporting season
The Commission built a standing unit for this category of case in February 2025 and has told the market that artificial intelligence misrepresentation is an immediate priority rather than a one time theme, so a company’s exposure is not limited to the small set of matters already charged. The Commission announced the Cyber and Emerging Technologies Unit on February 20, 2025, staffing it with approximately 30 fraud specialists and attorneys, with a stated mission that includes fraud committed using emerging technologies such as artificial intelligence and machine learning. The Commission’s acting chair at the time stated the unit exists to root out those who misuse innovation to harm investors, language the Commission has repeated in statements accompanying its 2025 and early 2026 matters.
A founder preparing a pitch deck or investor update in September 2026 is not drafting into a settled body of no action guidance. The Commission has stated publicly that it intends to keep bringing these cases, and every enforcement action charged to date turns on the same fact pattern: a present tense technical claim that internal records or a live demonstration contradict.
Want more industry insights? I publish daily analysis on the intersection of deep tech and venture law.
Open Questions
How far can a product roadmap describe a future AI capability before it becomes a present tense misrepresentation? The Presto and Nate matters both charged claims about current capability, and neither the Commission nor a court has drawn a bright line for a roadmap statement paired with a stated future date, leaving founders to guess at the wording that keeps a slide on the permitted side of the line.
Does a private company that never files periodic reports with the Commission face the same disclosure controls exposure that supported the Presto Automation charge? Presto’s Section 13(a) count depended on periodic reporting duties that apply only to registered issuers, so it is not yet clear whether a private company raising money under Regulation D faces exposure under Section 17(a) and Rule 10b-5 alone, and if so how a court will measure whether internal AI claim verification was adequate.
How will the Commission treat a founder who repeats a vendor’s inflated artificial intelligence marketing claim in good faith rather than fabricating one? Every matter charged to date involved a claim about the company’s own technology. The Raz and Saniger complaints do not address a founder who passes along an inaccurate claim a third party model vendor made first, a common pattern for companies built on a licensed model.
Will the Cyber and Emerging Technologies Unit bring cases against early stage companies that never close a round, or only against companies that actually raised capital on the strength of an AI claim? Every matter charged through 2026 involved completed raises measured in the tens of millions of US Dollars, leaving the smaller end of the venture market without a data point.
How does a change in Commission leadership affect the pace of this enforcement program? The current Commission chair has stated that its rules are principles based by design and not a prescriptive checklist for artificial intelligence disclosure, a description that leaves room for either a slower pace of new matters or continued reliance on the existing anti fraud statutes without new AI specific rulemaking.
Source List
1. United States Securities and Exchange Commission, Press Release No. 2024-36, SEC Charges Two Investment Advisers with Making False and Misleading Statements About Their Use of Artificial Intelligence (March 18, 2024), sec.gov. Primary source for the Delphia and Global Predictions penalties and statutory basis.
2. In the Matter of Delphia (USA) Inc., Investment Advisers Act Release No. 6573 (March 18, 2024), sec.gov. Primary source order text for the first AI washing settlement.
3. United States Securities and Exchange Commission, Order, In the Matter of Presto Automation Inc., Securities Act Release No. 33-11352 (January 14, 2025), sec.gov. Primary source for the negligence based AI washing theory against an operating company.
4. United States Securities and Exchange Commission, Press Release No. 2025-42, SEC Announces Cyber and Emerging Technologies Unit to Protect Retail Investors (February 20, 2025), sec.gov. Primary source for the Commission’s dedicated enforcement unit for artificial intelligence and cyber related fraud.
5. United States Securities and Exchange Commission, Press Release No. 2022-86, SEC Charges BNY Mellon Investment Adviser for Misstatements and Omissions Concerning ESG Considerations (May 23, 2022), sec.gov. Primary source for the greenwashing precedent the AI washing cases follow.
6. United States Attorney’s Office, District of Massachusetts, Former Technology Company Executives Charged with Defrauding Boston Private Equity Firm Out of 20 Million US Dollars, justice.gov. Primary source for a boundary case involving an application that did not work and fabricated financial records.
Bibliography
Primary Legal Sources
Securities Act of 1933 Section 17(a), 15 U.S.C. Section 77q(a).
Securities Exchange Act of 1934 Section 10(b), 15 U.S.C. Section 78j(b); 17 C.F.R. Section 240.10b-5.
Securities Exchange Act of 1934 Section 20(a), 15 U.S.C. Section 78t(a).
In the Matter of BNY Mellon Investment Adviser, SEC Press Release No. 2022-86 (May 23, 2022).
In the Matter of Delphia (USA) Inc., Investment Advisers Act Release No. 6573 (March 18, 2024).
In the Matter of Global Predictions Inc., SEC Press Release No. 2024-36 (March 18, 2024).
In the Matter of Goldman Sachs Asset Management, L.P., Investment Advisers Act Release (November 2022).
In the Matter of Presto Automation Inc., Securities Act Release No. 33-11352 (January 14, 2025).
Securities and Exchange Commission v. Albert Saniger, Litigation Release (S.D.N.Y. filed April 9, 2025).
Securities and Exchange Commission v. Illit Raz, Litigation Release (S.D.N.Y. filed June 11, 2024).
United States v. Cobb, Indictment (D. Mass. December 2025).
Secondary and Government Sources
United States Attorney’s Office, District of Massachusetts. Former Technology Company Executives Charged with Defrauding Boston Private Equity Firm Out of 20 Million US Dollars. Department of Justice, 2025. https://www.justice.gov/usao-ma/pr/former-technology-company-executives-charged-defrauding-boston-private-equity-firm-20.
United States Securities and Exchange Commission. SEC Announces Cyber and Emerging Technologies Unit to Protect Retail Investors. Press Release No. 2025-42, February 20, 2025. https://www.sec.gov/newsroom/press-releases/2025-42.
United States Securities and Exchange Commission. SEC Charges BNY Mellon Investment Adviser for Misstatements and Omissions Concerning ESG Considerations. Press Release No. 2022-86, May 23, 2022. https://www.sec.gov/newsroom/press-releases/2022-86.
United States 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. https://www.sec.gov/newsroom/press-releases/2024-36.
Interested in analysis about the intersection of tech, policy and the law? Check out my Substack channel. https://theinnovationattorney.substack.com/