AI Washing in RIA Marketing What Advisors Need to Know
By Ivan Barretto, RIA Compliance Concepts
AI Washing in RIA Marketing What Advisors Need to Know
AI Washing in RIA Marketing What Advisors Need to Know
Walk through almost any adviser’s website and you will see some version of the same promise: “AI-powered portfolios,” “machine-learning insights,” “intelligent planning,” or “automated intelligence for better outcomes.”
Some of those claims are accurate. Many are aspirational. A few describe technology the firm does not use at all.
Regulators have a name for the gap between what an adviser says about artificial intelligence and what the adviser actually does: AI washing. The term borrows from “greenwashing,” but the compliance issue is not new. It is the same old problem in a new wrapper. If a marketing claim is exaggerated, unsupported, or misleading, it can violate long-standing antifraud and advertising rules.
For RIAs, this is not just a technology issue. It is a Marketing Rule issue, a Form ADV issue, a fiduciary issue, and an exam readiness issue.
This article is for informational purposes only and is not legal advice.
AI washing means the claim outruns the reality
AI washing happens when a firm’s statements about artificial intelligence overstate, misdescribe, or fabricate how the firm uses the technology.
That can happen in obvious ways, such as claiming to use AI when no AI tool is involved. It can also happen in more subtle ways, such as using buzzwords that create a false impression about the role the technology plays.
Common examples include:
Calling a basic rules-based screen “AI”
Describing a third-party licensed tool as proprietary technology
Saying AI drives investment decisions when it only supports research
Claiming machine learning improves portfolio results without evidence
Suggesting a system is autonomous when humans make every final decision
Omitting known limits, data issues, or oversight weaknesses
Using AI-related claims in Form ADV that do not match actual operations
The core problem is not the use of artificial intelligence. Many advisers use AI tools in reasonable, useful ways. The problem is making a statement that a client, prospect, regulator, or examiner could reasonably read as more impressive than the facts support.
The words matter. So does the context.
“AI-assisted research” may be accurate if an analyst uses a third-party tool to summarize market data before making a recommendation. “AI-managed portfolios” may be misleading if no model actually selects, weights, or rebalances client holdings. “Proprietary AI engine” may be false if the firm simply licenses software from a vendor.
A useful test is simple: if an examiner asked the firm to prove the claim tomorrow, what would the file show?
The Marketing Rule already covers AI claims
AI washing does not require a special AI rule to become an enforcement risk. Existing adviser rules already cover the conduct.
Under the Investment Advisers Act of 1940, advisers owe fiduciary duties to clients. Section 206 prohibits fraudulent, deceptive, or manipulative conduct. Rule 206(4)-1, the SEC Marketing Rule, prohibits advertisements that include untrue statements of material fact or omit material facts needed to make statements not misleading.
The Marketing Rule also requires advisers to have a reasonable basis for believing they can substantiate material statements of fact on demand.
That point is critical. If a firm says it uses “machine-learning models to identify risk,” the compliance file should show what models exist, what they do, how they are used, who oversees them, and why the statement is fair. If the evidence is thin, the claim should be revised or removed.
Form ADV creates a second track of risk. The firm’s disclosure brochure, advisory business description, methods of analysis, risk disclosure, conflicts, and wrap fee or model descriptions must be accurate and not misleading. If marketing materials say one thing and Form ADV suggests another, that inconsistency may draw scrutiny.
State-registered advisers face similar issues under state securities laws. State regulators may not use the same language in every jurisdiction, but antifraud and advertising provisions often reach the same conduct.
The main point is clear: AI claims are advertising claims when they appear in websites, pitch materials, client communications, public profiles, consultant databases, webinars, videos, or other materials designed to attract or retain advisory clients.
Regulators are treating AI claims as an exam and enforcement priority
Regulators have moved from talking about AI washing to acting on it.
In recent years, the SEC has warned advisers and broker-dealers not to exaggerate their use of artificial intelligence. The agency has also brought settled enforcement actions against investment advisers over allegedly false and misleading AI-related claims. Those actions did not create a new duty. They applied familiar antifraud and advertising principles to a new set of claims.
The lesson for RIAs is practical. If a firm uses AI language to attract clients, the firm should expect exam staff to ask for support.
That request may not arrive as a dramatic “AI washing” inquiry. It may appear in an ordinary exam request list that asks for:
Marketing materials that reference AI, algorithms, machine learning, automation, or predictive analytics
Policies and procedures related to advertising review
Vendor due diligence files
Model governance documents
Data sources used by technology tools
Form ADV support
Client disclosures
Records showing how tools are used in the investment process
Supervision and testing records
The risk grows when AI language touches investment results, portfolio construction, risk controls, personalization, client profiling, or fee justification. Those claims can influence a client’s decision to hire or keep an adviser. That makes them material.
Most AI washing risk starts with imprecise language
A firm does not need bad intent to create AI washing risk. Often, the problem starts when marketing language runs ahead of the operating reality.
A portfolio manager may use a screening tool. A marketing writer may call it machine learning. A vendor may describe its product as AI-powered. A website may convert that into “our AI identifies opportunities.” By the time compliance reviews the page, the claim may sound polished, but the facts may be unclear.
Several patterns show up often.
A basic formula gets dressed up as artificial intelligence
Not every automated process is AI. A spreadsheet that ranks securities by valuation metrics is not artificial intelligence simply because it runs quickly. A rules-based rebalancing tool is not machine learning just because it acts without manual calculations.
If a process follows fixed instructions and does not learn from data, adapt, classify, generate content, or make predictions in a way commonly associated with AI, the safer language may be “automated,” “rules-based,” or “systematic.”
Those terms can still be valuable. They are also easier to prove.
A vendor tool becomes a proprietary claim
Many advisers use third-party software. That is not a problem. The risk appears when public statements imply the adviser built or owns technology that it only licenses.
A firm can usually say it uses a third-party tool, if that is accurate and consistent with vendor agreements. It should be careful about phrases such as “our proprietary AI,” “our exclusive model,” or “our in-house AI engine” unless the firm can prove those statements.
Vendor contracts may also restrict how the firm can describe the tool. Compliance review should include those limits.
A support tool sounds like a decision-maker
An adviser may use AI to summarize research, flag anomalies, draft internal notes, categorize service requests, or compare client data. Those uses may support human judgment. They may not drive advice.
That distinction matters.
If investment professionals make the final decision, say so. If the tool only provides inputs, say so. If the firm does not permit AI outputs to be used without human review, the marketing language should not imply autonomous decision-making.
Benefits are stated without support
Claims about better outcomes, greater accuracy, faster risk detection, improved personalization, or superior portfolio construction require evidence.
The Marketing Rule does not allow advisers to make material factual claims and hope they are true. The firm needs a reasonable basis. That may include testing records, model documentation, vendor materials, internal analyses, performance support, or other records that match the specific claim.
A general belief that AI is useful is not enough.
Limits and risks disappear from the message
Balanced disclosure matters. If a tool has known limits, the firm should be careful not to describe it in absolute terms.
AI tools can produce incorrect outputs. They may rely on incomplete data. They may reflect flawed assumptions. They may create privacy, confidentiality, cybersecurity, outsourcing, or recordkeeping concerns. They may work well for one task and poorly for another.
A marketing piece does not need to become a technical manual. Still, it should not create a false sense of certainty.
The main compliance question is whether the firm can prove the claim
Before an adviser publishes an AI-related statement, compliance should ask one question first:
What proof would show that this claim is accurate, fair, and not misleading?
That proof should exist before publication. Building the file after an exam begins is much harder.
The support should match the claim. A high-level vendor brochure may support a statement that the firm licenses a tool with certain features. It may not support a claim that the adviser’s use of the tool improves client outcomes. Internal notes may show that analysts use AI for research summaries. They may not support a claim that an algorithm manages client portfolios.
Here is a practical way to think about common claims.
Marketing claim | Support the firm should have | Safer wording if support is limited |
“We use AI to manage portfolios” | Documentation showing AI is part of portfolio construction, trading, rebalancing, or allocation decisions, plus oversight records | “We use technology tools to support portfolio research and monitoring” |
“Our proprietary AI model identifies opportunities” | Ownership records, model documentation, testing, governance, and evidence showing what the model identifies | “We use internal screening tools to help review potential investments” |
“Machine learning improves risk management” | Testing, methodology, performance support, and records tying the tool to risk processes | “We use data tools to help monitor selected risk factors” |
“AI creates personalized financial plans” | Documentation showing AI generates plan elements and how humans review them | “Planning software helps organize client data and scenarios for adviser review” |
“Our AI reduces errors” | Valid testing comparing error rates before and after use | “Automation helps standardize parts of our review process” |
The safest claims tend to be specific, narrow, and tied to actual use. The riskiest claims tend to be broad, vague, and impressive sounding.
A strong review process starts with an AI inventory
An adviser cannot review AI claims well unless it knows where AI is used. The first step is a plain-English inventory.
The inventory should cover more than investment models. AI may appear in several parts of the business:
Portfolio research
Trading or rebalancing support
Financial planning tools
Client risk questionnaires
Note-taking or meeting summaries
Client service chat tools
Email drafting or review tools
Compliance surveillance
Data extraction
Cybersecurity monitoring
Vendor platforms with embedded AI features
For each tool, the firm should document what it is, who owns it, who uses it, what data it touches, what outputs it creates, and how humans oversee it.
This inventory does not need to be elegant. It needs to be usable. Compliance, operations, investment, technology, and senior management should agree on what the tools do. If the people inside the firm cannot describe the technology consistently, marketing should not describe it publicly.
Policies should connect technology use to marketing review
Many firms have advertising policies. Many also have technology, cybersecurity, vendor, privacy, and portfolio management policies. AI claims often sit at the intersection of all of them.
A practical policy should state that any public or client-facing reference to artificial intelligence, machine learning, algorithms, automation, predictive analytics, or similar technology requires review before use.
The review should check several points.
Accuracy
Does the statement match actual firm practice?
Substantiation
Can the firm prove each material statement of fact?
Context
Would a reasonable client misunderstand the role of the technology?
Ownership
Does the language imply the firm built, owns, or controls a tool when it does not?
Human oversight
Does the statement explain the role of adviser judgment when that role is material?
Risks and limits
Does the piece omit facts needed to make the claim fair?
Consistency
Does the statement match Form ADV, client agreements, vendor agreements, policies, and actual workflows?
These questions should apply before publication, not after a concern arises.
Form ADV should match the marketing story
If a firm’s website claims that AI plays a central role in portfolio management, Form ADV should not read as if the firm uses only traditional discretionary management with no technology component. If Form ADV describes a quantitative model, marketing should not add unsupported AI claims that go beyond the brochure.
Key areas to review include:
Advisory business descriptions
Methods of analysis
Investment strategies
Material risks
Disciplinary and conflict disclosures, if relevant
Wrap fee or model portfolio descriptions
Other financial industry activities and affiliations, if vendors or affiliates are involved
Brochure supplements, where individual supervision or role descriptions may matter
Form ADV does not need to repeat every marketing phrase. It does need to avoid inconsistency and omission. If artificial intelligence is material to the advisory service, the brochure should describe it accurately. If AI is not material, marketing should not make it sound central.
Vendor diligence matters because vendor language can become your claim
Vendors often use ambitious descriptions of their products. That language may be useful for vendor sales, but it is not automatically safe for an RIA’s client-facing materials.
Before repeating vendor claims, the adviser should understand what the vendor can support. Useful records may include:
Product descriptions
Technical documentation at a level the firm can understand
Testing or validation materials
Data source information
Security and privacy documentation
Service agreements
Usage limits
Disclosures and disclaimers
Contract language governing public references
Records showing which features the adviser actually uses
The last point matters. A vendor platform may contain AI features that the adviser has not enabled. A firm should not claim to use those features just because they exist in the product suite.
Vendor diligence should also address data. If a tool uses client information, the firm should understand whether data is stored, shared, retained, used for model training, or processed by subcontractors. Those facts may affect privacy disclosures, cybersecurity controls, client communications, and risk assessments.
Better wording can reduce risk without hiding useful technology
Compliance review should not force every firm to avoid AI language. If an adviser uses artificial intelligence in a real and meaningful way, it can say so. The goal is to make the statement accurate and fair.
Often, the fix is not silence. It is precision.
Less precise | More precise |
“Our AI manages your portfolio” | “Our investment team uses data analysis tools to support portfolio monitoring, and advisers make final portfolio decisions” |
“AI-powered planning gives clients better outcomes” | “Planning software helps us compare scenarios and organize client information for adviser review” |
“Our proprietary machine-learning platform finds hidden risks” | “We use a licensed analytics platform to help flag selected risk factors for further review” |
“Intelligent algorithms remove emotion from investing” | “Rules-based portfolio guidelines help support consistent review and rebalancing decisions” |
“AI gives us superior market insight” | “Research tools help summarize market data, which our investment team reviews as part of its process” |
This kind of wording may sound less dramatic. It is also easier to defend.
Clients do not need inflated technology claims. They need to understand how the adviser works, what tools support the service, and where human judgment remains involved.
The compliance file should tell the same story as the website
An examiner reviewing AI-related claims will not look only at the final advertisement. The examiner may compare the advertisement to internal records.
A strong file may include:
Final approved marketing materials
Drafts showing compliance comments
Substantiation for each material AI claim
AI tool inventory
Vendor diligence
Relevant contracts
Model or tool governance records
Testing or validation records, if claims depend on them
Evidence of human review and supervision
Related policies and procedures
Training records
Form ADV support
Records of periodic reviews
The firm should also keep records showing that it removed or revised outdated claims. AI tools change quickly. A statement that was accurate last year may become stale if the vendor changes the product, the firm stops using a feature, or the firm’s process changes.
Marketing review should not be a one-time event. It should be part of the compliance calendar.
Training helps prevent small wording problems from becoming firm problems
AI washing risk often begins outside the compliance department. A portfolio manager speaks at an event. A client service team updates a description. A marketing contractor adds buzzwords. A vendor provides suggested copy. A partner edits a pitchbook before a prospect meeting.
People do not need to become AI engineers to avoid the problem. They need a short list of rules.
A useful training message might include:
Do not call a tool AI unless the firm has approved that description
Do not describe vendor technology as proprietary
Do not claim improved results without approved support
Do not imply the tool makes decisions if humans decide
Do not copy vendor AI language into client materials without review
Send new or revised technology claims to compliance before use
Training should also explain why the issue matters. The concern is not whether a phrase sounds modern. The concern is whether the phrase could mislead a client or prospect.
Exam readiness means having answers before questions arrive
An RIA that uses AI language should be ready for direct questions. The best answers are factual, short, and supported by records.
Compliance and management should be able to explain:
Which AI tools the firm uses
Which tools are client-facing
Which tools affect investment advice
Which tools use client data
Who approves use of each tool
How outputs are reviewed
What vendors provide the tools
What claims the firm makes publicly
What evidence supports those claims
How the firm monitors changes over time
If the answer to several of those questions is unclear, the firm should pause broad AI claims until the facts are organized.
A simple internal review can help. Search the firm’s website, pitchbooks, Form ADV, brochures, client letters, biographies, videos, articles, and consultant profiles for terms such as:
Artificial intelligence
AI
Machine learning
Algorithm
Predictive
Automated
Intelligent
Data-driven
Quantitative
Proprietary technology
Not every use of those terms is risky. Each one deserves context.
AI can be real and still be poorly described
One of the harder issues is that a firm may genuinely use AI and still create a misleading impression.
For example, an adviser may use an AI tool to summarize earnings call transcripts. That is a real use. But if the firm says “AI identifies securities for client portfolios,” the statement may go too far.
Another adviser may use machine learning in a risk model. That may be meaningful. But if the model covers only a small portion of the process, marketing should not suggest that AI controls the full portfolio.
A third firm may use generative AI to draft planning summaries. That may save time. But it does not mean the firm provides AI-generated financial advice, especially if advisers review and edit every output.
Accuracy is not only about whether AI exists. It is about what AI does, how much it matters, and what role people play.
Treat AI claims like performance claims
Performance advertising already teaches a useful lesson. Advisers know they should not publish return claims, rankings, or comparisons without support, context, and required disclosures. AI claims deserve the same discipline.
Before publishing, ask:
Is the claim factual or promotional puffery?
Would a reasonable person view it as material?
Can the firm prove it?
Does the proof match the exact wording?
Does the statement need limits or context?
Is the claim still true today?
Would Form ADV, vendor files, and internal workflows support it?
If the answer is uncertain, revise the language.
The best compliance posture is not anti-technology. It is pro-accuracy. Advisers can use artificial intelligence, automation, and data tools in valuable ways. They just need to describe those tools with the same care they bring to fees, conflicts, performance, and investment process.
The practical takeaway for RIAs
AI washing in RIA marketing is not a passing vocabulary problem. It is a legal and compliance problem created when public claims move faster than facts.
The fix is straightforward, but it takes discipline:
Inventory actual AI use
Review every public claim
Match each claim to support
Align marketing with Form ADV
Check vendor language before repeating it
Explain human oversight clearly
Keep records that prove the review happened
Revisit claims as tools and workflows change
Advisers do not need to avoid technology language. They need to earn it.
If a firm can prove what its AI does, describe the limits, and show how people supervise the process, the marketing can be both compelling and compliant. If the proof is not there, the better choice is to say less, say it plainly, and fix the process before the claim becomes an exam issue.
Comments