SEC Brings First AI-Washing Enforcement Action Against a Mid-Cap Issuer

sec securities disclosure November 14, 2025 · by Eleanor Hartwell

The SEC announced a settled administrative proceeding on November 6 against Lucentra Software Inc., a mid-cap enterprise SaaS company, charging the company with making materially false and misleading statements about its AI capabilities in earnings calls, press releases, and 10-K filings during fiscal years 2023 and 2024. Lucentra agreed to a $14.5 million civil penalty, neither admitting nor denying the findings.

This is the first significant AI-washing case to settle. The SEC has had several investment-adviser-side AI-washing actions in 2024, and one smaller issuer case earlier this year, but Lucentra is the first major issuer enforcement. The order will be widely cited and is worth careful reading for anyone advising public companies on AI disclosure.

What Lucentra allegedly did

The order describes four categories of alleged misstatement:

1. Capability overstatement. Lucentra's executives stated, on multiple earnings calls, that the company's flagship product was "powered by proprietary large language model technology developed in-house." The technology was, in fact, primarily a wrapper around third-party API calls to commercial models, with limited proprietary fine-tuning on top.
2. Customer-adoption inflation. The company reported "AI-product adoption" metrics in supplemental disclosures that included customers who had merely been exposed to AI features in marketing without using them, and customers using a beta product that the company described as "in production."
3. Margin misrepresentation. The company described AI-product margins in earnings calls as "consistent with our broader software margins" when, in fact, the API-call costs from third-party providers materially compressed those margins. The order suggests the actual margins were materially lower.
4. R&D spending characterization. The company's 10-K disclosures categorized substantial research and development spend as "AI-related," when much of that spend was on infrastructure and product engineering with attenuated AI relevance.

Cumulatively, the SEC's case is that Lucentra inflated the perceived strategic positioning of its AI capabilities for the purpose of supporting its stock price. The order references analyst reports that explicitly quoted Lucentra's overstatements as supporting their valuations.

The legal theories

The order charges:

• Section 10(b) of the Exchange Act and Rule 10b-5: material misstatements in statements made to the public.
• Section 17(a) of the Securities Act: misstatements in connection with the sale of securities.
• Sections 13(a) and 13(b) of the Exchange Act and related rules: books-and-records and internal controls violations stemming from the customer-adoption metric methodology.

The 10b-5 theory is the most interesting. The order articulates a clean statement of when AI capability claims are material: where the issuer's stock price is meaningfully driven by perceptions of AI capability, public statements about that capability are presumptively material, and reasonable investors would rely on them. This is a more demanding materiality formulation than I expected.

The "puffery" defense, often available for forward-looking AI capability statements, was not available here because the statements were specific and verifiable: "developed in-house," "in production," "consistent with broader margins." Each of these is a fact-specific representation that can be falsified.

Internal-controls implications

The Section 13(b) charges deserve specific attention. The SEC alleges that Lucentra lacked adequate internal controls to ensure that the customer-adoption metrics it reported were calculated according to consistent and disclosed methodologies. Several internal documents — produced through the investigation — show the methodology for these metrics shifting between quarters in ways that flattered reported numbers.

For practitioners advising public-company disclosure committees, this is the durable lesson. AI-related metrics are now being reported widely in supplemental disclosures, MD&A, and earnings calls. The internal controls around how those metrics are defined, calculated, and validated need to be at least as robust as the internal controls around financial-statement metrics. They generally are not.

What the SEC has signaled

The order is unusually long and unusually explicit about the SEC's general posture. Three things to read out of it:

1. The SEC is going to bring more AI-washing cases. The order expressly references "the increasing prevalence of issuer AI-related disclosure" and "the importance of accurate disclosure to capital formation." This is signaling, not just adjudication.
2. Disclosure committees should be running AI-claim review processes. The order describes Lucentra's failure to put AI-capability statements through the same review processes used for financial-statement disclosures. This is a common pattern across issuers, and it is going to be the next compliance frontier.
3. External advisor guidance is on the SEC's mind. The order mentions, in passing, that Lucentra's auditors and outside counsel raised questions about the AI-product margin claims that were not adequately addressed by management. The SEC will use cases like this to draw audit firms and law firms further into AI-disclosure scrutiny.

What public companies should do

For public-company AI disclosure programs:

• Run an AI-claim audit. Pull the past four quarters of earnings calls, press releases, and SEC filings. Identify every public statement about AI capability, AI product adoption, or AI-related financial performance. For each, identify the supporting evidence and verify against current reality. The exercise is uncomfortable but informative.
• Disclosure committee processes should explicitly cover AI claims. This means having a person in the room who understands the technology well enough to challenge characterizations, not just review them.
• Define AI metrics precisely and disclose the definitions. "AI-product adoption" can mean many things; whatever it means in your filings should be defined and applied consistently.
• Margin and unit-economics disclosures involving third-party API costs need careful treatment. The compressed-margin reality of API-wrapped products is a difficult disclosure conversation, but failing to have it creates the Lucentra problem.
• For private companies preparing for IPOs: get this right before the prospectus rather than after.

The broader signal

The SEC's pivot toward AI-disclosure enforcement is part of a broader pattern of agencies that survived the federal AI-policy pivot finding their footing on AI-related issues through their existing authorities. The CFPB has continued AI-related fair-lending work; the FTC has continued AI-related deceptive-practices work; the SEC has now staked out securities-disclosure territory. The cumulative regulatory pressure on public companies handling AI claims is meaningful even without any new AI-specific federal statute.

The Lucentra order is going to be the watershed reference. Read it now.