AI Washing: The Marketing-Rule Trap RIAs Are Walking Into

Last updated: 2026-08-25 · By Akhil Lodha

AI washing — overstating or misrepresenting how a firm uses artificial intelligence — is already an enforcement category, not a theoretical risk. The SEC has charged investment advisers for false AI claims under existing antifraud and marketing provisions, and every "AI-powered" phrase on an RIA's website is now a claim the firm must be able to substantiate in an exam.

The trap isn't lying about AI. It's rounding up.

Nobody at your firm sat down and decided to deceive prospects about artificial intelligence. That's what makes AI washing dangerous. The trap is gradual: a vendor's AI feature becomes "our AI-driven process" on a pitch deck. A chatbot the ops team uses becomes "we harness machine learning" on the website. Marketing rounds up, because marketing always rounds up — and the SEC reads the rounded-up version literally.

Here's the reframe: the SEC didn't create a new AI marketing rule, and it didn't need to. AI washing is a plain-vanilla misleading-statement problem wearing new clothes. Rule 206(4)-1 — the Marketing Rule — prohibits advertisements containing untrue statements of material fact, or statements the adviser cannot substantiate upon SEC demand. "Cannot substantiate upon demand" is the phrase that should be circled in red. It shifts the burden to you: say "AI-powered," and you must be able to show, on request, exactly what that means.

AI washing is the practice of making claims about a firm's use of artificial intelligence that are false, exaggerated, or unsupportable — the AI-era descendant of greenwashing. For an RIA, it lives at the intersection of the Marketing Rule, the general antifraud provisions of the Advisers Act, and the SEC's stated examination focus on advisers' use of AI in its FY2026 exam priorities.

What has the SEC actually done about AI washing?

The founding precedent came fast. In March 2024, the SEC charged two investment advisers — Delphia (USA) Inc. and Global Predictions Inc. — with making false and misleading statements about their purported use of artificial intelligence, settling for $400,000 in combined penalties, per SEC press release 2024-36. Delphia had advertised that its algorithm used client data to predict winning companies and trends before the rest of the market — capabilities the SEC's order found it never actually had. Global Predictions marketed itself as the "first regulated AI financial advisor," a claim it could not substantiate when asked.

Read those cases carefully and the lesson is uncomfortable: these weren't firms with no AI. They were firms whose _descriptions_ of their AI didn't match reality. The gap between what the technology did and what the marketing said — that gap was the violation. Then-Chair Gensler framed it as classic securities law: firms talking up their AI must make sure what they tell investors is true. And there's a detail most coverage skipped: both firms were also charged under Rule 206(4)-7 for failing to have policies reasonably designed to prevent these misstatements — meaning the SEC's first AI-washing cases doubled as compliance-program cases. The scrutiny hasn't faded with the change in administration, either: the FY2026 exam priorities, the first issued under Chairman Atkins, keep advisers' use and representation of AI squarely in focus.

Which claims put an RIA at risk?

The risky claims are rarely exotic. They're the sentences that feel harmless in a website review:

Two patterns account for most exposure. Vendor inflation: describing a third-party tool's AI as the firm's own capability. Using AI-enabled software is fine; claiming "our AI" when it's a subscription is a misstatement waiting for a request letter. Aspiration as fact: describing the roadmap in the present tense. "We use AI to detect risk drift" must be true on the day it's published, not by Q4.

What does an AI-washing finding actually cost?

The penalty is the smallest line item. The Delphia and Global Predictions settlements were $225,000 and $175,000 — survivable numbers. What isn't survivable so easily: a public order describing your firm as having misled clients, sitting at the top of every prospect's search results; an amended Form ADV process; and the downstream exam attention that a marketing finding invites, because an examiner who finds one unsupported claim re-reads everything else with new skepticism. For a firm whose entire business is fiduciary trust, "the SEC said we misled people about our technology" is a sentence sales never fully recovers from — every prospect who searches your firm's name finds it before they find your pitch.

How do you market AI honestly — without unilaterally disarming?

The answer isn't to stop talking about AI. Firms that genuinely automate document processing or portfolio supervision have a real story, and prospects want it. The answer is claim discipline:

1. Inventory every AI claim. Website, decks, ADV Part 2, one-pagers, LinkedIn. If the word "AI" (or "machine learning," "intelligent," "smart") appears, it goes on the list.
2. Attach substantiation to each claim. For every sentence, a file: what tool, what it does, whose it is, evidence it's in production. If the file can't be built, the sentence gets rewritten.
3. Describe the mechanism, not the magic. "Our platform reads uploaded account statements and extracts holdings automatically, which our team then reviews" is substantiated, specific — and more persuasive than "AI-powered onboarding."
4. Route AI claims through compliance review, and log it. Marketing-claim review is supervision, and supervision needs an audit trail. A logged review of each claim, with its substantiation, is exactly the artifact that ends an exam inquiry quickly.

That last step is where process meets platform. Firms running supervision through StratiFi's ComplianceIQ get the review trail automatically — approvals and documentation created as a byproduct of the workflow, ready when the request letter arrives. And there's a quieter benefit: when your firm can _show_ what its AI does — statements read by AdvisorIQ · Statement Scanning, exceptions surfaced in OperationsIQ, oversight evidenced in ComplianceIQ — you don't need inflated language. The demo substantiates the claim. For where AI representations sit in the current exam cycle, see our SEC 2026 exam priorities analysis, and for the governance program that should sit behind your claims, our guide to the AI governance questions examiners are asking.

Frequently Asked Questions

If you want the substantiation problem to solve itself — real AI doing documented work, with the audit trail built in — subscribe to the newsletter for the ongoing AI-compliance series, or book a demo and see what defensible "AI-powered" actually looks like.