With increasing use of artificial intelligence (“AI”), many private fund managers have turned to proprietary AI algorithms as the basis for their investment strategies. While these AI algorithms offer the potential for significant returns, many fund managers and issuers routinely overstate their AI capabilities in pitch decks and private placement memorandums. “AI washing” refers to when issuers make false, misleading, or exaggerated claims about their AI capabilities. Recently AI washing has become an area that the Federal Trade Commission “FTC”, Securities and Exchange Commission “SEC”, and other regulatory agencies are cracking down on. Starting in early 2024, the SEC has brought multiple actions against issuers and investment advisers for AI washing, established a dedicated enforcement unit, and stated that it will use existing antifraud rules to regulate the space. This article explains the legal framework, identifies patterns that get issuers in trouble, and offers practical guidance for companies raising capital.
I. LEGAL FRAMEWORK
Regulation D, Rules 506(b) and 506(c) of the Securities Act of 1933 exempt certain offerings from registration requirements. However, while Regulation D offerings aren’t required to file a registration statement and undergo SEC review, the antifraud provisions of the federal securities laws still apply, meaning that under SEC guidance, any information provided to investors must be free of material misstatements and must not omit facts necessary to prevent other statements from being misleading (See Rule 10b-5 of the Securities Exchange Act of 1934 and Section 17(a) of the Securities Act).
A misstatement or omission is material if a reasonable investor would consider the information significant in deciding whether to invest. For issuers whose investment strategy is based on AI or their proprietary AI algorithm, this standard is typically easily met. However, notably, puffery is different from a material misstatement. An issuer saying that their company is at the forefront of the AI revolution would likely be deemed puffery and therefore not as problematic as an issuer stating that their platform uses proprietary AI to autonomously complete every transaction when that is not actually the case and is therefore a misstatement. As an example, in 2024, Delphia was charged for claiming that it put collective data to work to make its artificial intelligence smarter so it can predict which companies and trends are about to make it big and invest in them before everyone else. The SEC found that these statements were false and misleading because Delphia did not actually have the AI and machine learning capabilities claimed.
II. WHERE PRIVATE ISSUERS GET INTO TROUBLE
a. Overstating Capability
The AI washing pattern that is the most common and most aggressively pursued by the SEC is where issuers claim that AI performs functions that actually require human intervention. Words such as “fully automated” or “AI powered end to end processing” are therefore very problematic when human review, corrections, or completion is required for tasks that AI was alleged to handle. Even if the issuer aspires to eventually reach a place where zero human intervention is required, the issuer needs to be honest about the AI’s current capabilities and the role of the technology versus human involvement.
b. Misattributing Technology
Another common problem is when issuers represent that AI is proprietary when it is actually built on top of foundation models from sources such as OpenAI, Anthropic, Google, or others. When the underlying technology is licensed, issuers stating that the resulting product is “proprietary AI” is problematic because it creates a false impression about the company’s competitive position. As an example, the SEC charged Presto Automation finding that its statements were misleading since its technology was owned and operated by a third party. Because of this, issuers need to ensure that their pitch decks and offering documents accurately describe what the company built independently, what it licenses, and any risks related to third party dependencies.
c. Inconsistency Across Documents
The third common problem is inconsistency between the pitch deck and private placement memorandum. Issuers should ensure that both documents tell the same story and that messaging is consistent across the board. It is very common for the pitch deck as a marketing document to overstate AI’s capabilities which can result in misstatements to investors and scrutiny by securities regulators.
III. PRACTICAL GUIDANCE
When describing your offering, investment strategy, and technology, it is imperative for issuers to be clear and accurate in describing the technology’s current capabilities, disclosing any third party dependency, describing human involvement honestly, and avoiding exaggerating performance (i.e., “best in class” or “industry leading”) without substantive data to back it up. Issuers should also ensure that marketing materials and the offering documents align and tell the same story when it comes to the AI technology. Additionally, AI products change quickly. Issuers should ensure that the offering documents and marketing materials accurately reflect the technology’s current status with updates as the AI technology changes.
The Cost of Getting This Wrong
AI washing is an area where regulatory agencies have shown greater interest in enforcement actions and where we can predict greater scrutiny in the future. Potential liability includes both criminal prosecution and civil penalties. Therefore, it is imperative to treat AI capability statements with the same discipline applied to financial projections, substantiate statements before releasing them, ensure consistency across documents, update disclosures when things change, and describe the human versus technological role.
If you have questions about aligning your disclosures, marketing materials, and technology representations, or would like guidance on avoiding AI washing, please reach out to our team, and we’d be happy to assist.