What a Criminal Wire Fraud Charge Over "AI" Teaches Every Fintech Founder

AI-related securities class action filings more than doubled in a single year, from seven cases in 2023 to fifteen in 2024, according to Cornerstone Research and Stanford Law School . By 2025, that number kept climbing. In April 2025, one of those cases got a face. The SEC and DOJ charged Albert Saniger , founder of the shopping app Nate Inc., with wire fraud for claiming Nate's app used AI to process transactions. It didn't. Manual workers completed the purchases. He told investors the automation rate was above 90 percent. The actual rate was close to zero. He had raised more than $42 million on that claim. This is not a hypothetical regulatory risk. A founder is facing criminal charges right now for describing his product's AI capabilities inaccurately, and fintech is exactly where this enforcement trend is concentrated. What AI-washing means in practice The SEC's own term for it is straightforward: a company claims to use AI to enhance its products or services when, in fact, it does not, or the claim materially overstates what the AI really does. The first enforcement actions came in March 2024, against two investment advisers, Delphia and Global Predictions . Delphia had told clients for three years that it used AI and machine learning to analyze their personal data, things like social media, banking, and credit card activity, as inputs into its investment algorithms. It never did. Both firms settled, paying $225,000 and $175,000 in penalties. That was the template. Since then, the enforcement pattern has expanded well past investment advisers. The shift founders should be watching The most important change is not the number of cases. It is who the cases are now targeting. In January 2025, the SEC charged Presto Automation , a Nasdaq-listed company, marking the first AI-washing action against a public operating company rather than an investment adviser. Presto had marketed its drive-thru ordering product as eliminating the need for human order taking. The SEC found the vast majority of orders still required a human. Presto had also failed to disclose that the speech recognition technology powering the product was owned and operated by a third party, not built in-house. Then the Nate case moved the exposure from civil penalties to criminal charges. Saniger was not just fined. He was indicted. The direction is clear. Enforcement started with investment advisers making misleading claims to clients. It has moved to operating companies making misleading claims to investors, users, and the market. Fintech products sit directly in that path, because AI claims in fintech marketing - fraud detection, underwriting, KYC automation, portfolio management - are exactly the kind of claims regulators are now testing against what the engineering really does. The pattern behind every case that gets charged Look closely at the cases that got charged, and a consistent pattern emerges. It is rarely a single false statement. It is usually one of a few structural gaps. Claiming proprietary technology that is really third-party. Presto marketed its AI as its own; the underlying technology belonged to someone else. DocGo Inc. faced a March 2025 ruling allowing investor claims to proceed over a proprietary central AI system that did not function as represented. Overstating the automation rate. Presto claimed no human intervention was needed. Nate claimed automation above 90 percent. In both cases, the real number was close to the opposite of the public claim. Claiming AI is doing something it was never doing at all. Delphia claimed client data fed an AI-driven investment algorithm. The data collection happened. The algorithm never used it. Every one of these gaps is checkable. The SEC's approach has been to compare public representations against actual system performance data, pulling logs, architecture documentation, and vendor contracts to see whether the marketing matches the build. I think of this as the claims-to-code gap: the distance between what your pitch deck, your website, and your investor updates say your AI does, and what your actual codebase can prove it does. Every AI-washing case charged so far is, underneath the legal language, a claims-to-code gap that someone eventually measured. What closing the gap looks like in practice This is not a legal exercise you run once before a fundraise. It is a standing discipline, and it maps to three concrete checks. First, a claims register. Every public statement about what your AI does, pulled from your website, your deck, your app store listing, your investor updates, gets logged against the specific feature it describes. Not a general impression. A named claim, tied to a named capability. Second, a capability match. Each claim in that register gets checked against what the system really does in production right now, not what it was scoped to do, not what is on the roadmap. If your deck says AI-powered fraud detection and what is shipped is a rules engine with a machine learning model still in shadow mode, that is the exact gap Presto and Nate got charged over. Third, a provenance flag. For every AI capability you claim, is it built in-house or licensed from a vendor? This single distinction is the most frequently repeated fact pattern across all cases charged so far. If it is licensed, your claim needs to say so. None of this requires a legal team to run. It requires someone who can read both your investor deck and your architecture diagram and tell you honestly where they diverge. Why this is now a standing priority, not a one-off sweep In February 2025, the SEC created the Cyber and Emerging Technologies Unit , roughly 30 fraud specialists and attorneys, with fraud involving AI and machine learning named as its first priority area. This is not a temporary sweep tied to one administration's agenda. The unit exists specifically to keep testing AI claims against reality, on an ongoing basis. The SEC has also been explicit that it does not believe new AI-specific rules are necessary. Existing anti-fraud and disclosure laws already cover misleading AI claims. That matters for founders: there is no rulebook to wait for. The exposure already exists under the securities laws you are already subject to. And even where SEC enforcement volume shifts with a change in administration, the private litigation numbers tell their own story. AI-related securities filings did not slow down between 2023 and 2025. They kept climbing. The question worth sitting with What does your claims register look like right now, in reality: current, informal, or nonexistent? Founders who can answer that question quickly are the ones least exposed here. This is a general overview of a developing enforcement area, not legal advice. If your AI-related disclosures are being questioned, or you want a second set of eyes before they are, talk to securities counsel. I write about AI, fintech, and the gap between what founders claim and what their systems can prove. If this was useful, let's connect on LinkedIn .