Alex ZonneveldThe most useful AI in finance is usually the least visible, and the AI that’s advertised most loudly is often doing the least.
Almost every wealth platform now says it uses AI, and some of those claims are much stronger than the products behind them. Regulators have a name for overstating it, AI washing, and they’ve brought cases against firms that described manual processes or simple rules as artificial intelligence. The useful question for a capital owner isn’t whether a product uses AI. It’s what the AI actually does, what data it can see, what it’s allowed to do with your capital, where its numbers come from, who checks it and what happens when it’s wrong. AI that reads, reconciles, monitors and explains tends to be genuinely valuable, while AI that claims to make better decisions with your money deserves the most scrutiny.
AI washing is now a regulatory issue
In March 2024, the SEC brought its first AI-washing cases against two investment advisers, Delphia and Global Predictions, which together paid $400,000 in penalties. The SEC found that Delphia had claimed to use AI and machine learning on client data in ways it didn’t, and that Global Predictions had described itself as the “first regulated AI financial advisor”. In January 2025 the SEC settled with Presto Automation, a restaurant technology company, over claims about an AI voice ordering product that relied heavily on human workers. In April 2025, federal prosecutors and the SEC charged Albert Saniger, the founder of a shopping app called Nate, alleging that he raised more than $40 million by telling backers the app used AI to complete purchases when contract workers in call centres were doing much of the work.
The regulators have kept the issue in view since. The SEC’s 2026 examination priorities, published in November 2025, said examiners would review the accuracy of firms’ representations about their AI capabilities, and its 2027 priorities, published in October 2026, give AI even more attention. For anyone choosing a platform, the practical lesson is that “powered by AI” is a claim worth testing.
Two kinds of AI theatre
The first kind is overclaiming, where AI is described as making better decisions than people can, finding returns others miss or managing your money intelligently, with little evidence to support it. That should make anyone cautious, because when leading AI models were given real money to trade in public contests, most of them lost it, as we describe in AI is an accelerant for specialists, not a substitute for them.
The second kind is decoration, where a chat window or a set of automated summaries sits on top of a product that works exactly as it did before. That isn’t necessarily harmful, and it can be convenient, but it doesn’t change the things that decide whether capital is managed well, which are risk, liquidity and access.
The AI that genuinely helps is often much quieter. It reads offering documents and extracts the terms that matter, reconciles a manager’s reported numbers against the custodian’s records, monitors drawdowns and redemption queues, and explains what’s happening in plain language. None of that makes for a dramatic headline, but all of it is real work that used to require people.
Six questions that separate the two
The first question is what the AI actually does, described in plain verbs. If a provider can tell you that it reads your fund documents and extracts the liquidity terms, or that it reconciles balances daily and alerts you to differences, you’re dealing with something concrete. If the answer is that it optimises your portfolio intelligently, ask what that means in practice.
The second is what data it can see, because connecting an AI to your accounts gives it a detailed view of your financial life, so it’s worth knowing exactly which data it can access, whether that data is used to train models or shared with anyone else, and how you can revoke access. McKinsey has described permissioned data as one of the control points that will decide where value sits in AI-driven wealth management, and it matters just as much to the owner of the data.
The third is what it’s allowed to do, since an AI that can only read and report carries very different risks from one that can trade, and one that can trade is very different from one that can move assets. That’s the subject of Permissioned agents, and it’s the question that matters most when something goes wrong.
The fourth is where its numbers come from, because language models can state figures with complete confidence whether or not they’re accurate, so any number that matters should be traceable to a source, such as the custodian, the trading venue or an independent administrator, rather than generated by the model.
The fifth is who checks it, and the SEC has said it will look at firms’ governance and human oversight of AI, so capital owners should ask the same thing: who reviews what the AI produces, how it’s tested, and how often.
The sixth is what happens when it’s wrong, because models can give different answers to the same question and can be manipulated by content they read, so it’s worth knowing whether its outputs are logged, whether mistakes can be traced and corrected, and who is accountable for the consequences.
Red flags worth noticing
A few signals deserve particular caution, and one of them is any suggestion that AI will produce better returns, especially without a long, independently verified record. Another is an AI that chooses strategies or managers for you rather than helping you evaluate them, and another is figures that appear without any indication of where they came from. Requests for broad access to your accounts or authority to move assets are a warning sign, and so are explanations that rely on the word “AI” rather than describing what the system does. If a provider can’t explain its AI in a few plain sentences, it’s reasonable to wonder how much of it there is.
How MakeBanc approaches it
MakeBanc tries to hold itself to the same standard it suggests for others, by describing what its infrastructure does rather than leaning on the word AI. Your assets sit with an institutional custodian in a vault you sign for, the manager you choose can trade but never move them, every number is labelled as self-reported, observed or verified, NAV is calculated by a third-party administrator and reconciled daily, and the liquidity terms and costs of a strategy are visible before you allocate. MakeBanc doesn’t use AI to choose strategies for you, to make allocation decisions or to move your assets, and those decisions stay with you. Its agent, when it launches, doesn’t trade: it follows the allocation rules you set, and you sign every move.
Frequently asked questions
What is AI washing? AI washing is overstating or misrepresenting the use of artificial intelligence in a product or service. The SEC has brought cases against firms including Delphia and Global Predictions in 2024 and Presto Automation in 2025, and has said it will review the accuracy of firms’ claims about their AI capabilities.
How do I evaluate an AI wealth management tool? Ask what the AI actually does in plain terms, what data it can see and how that data is used, what it’s allowed to do with your capital, where its numbers come from, who checks its outputs, and what happens when it’s wrong.
What are the red flags in AI wealth management products? Claims that AI will produce better returns without a verified record, AI that chooses strategies for you, figures without a stated source, requests for broad account access or authority to move assets, and explanations that rely on the word AI rather than describing what the system does.
What kind of AI is genuinely useful for managing money? AI that reads documents and extracts key terms, reconciles reported figures against custodian records, monitors risk and liquidity, and explains what’s happening in plain language, while leaving decisions and control of assets with the owner.
Sources
- Holland & Knight, Beyond the hype: the SEC’s intensified focus on AI washing practices, April 2024
- Insurance Journal, SEC charges two investment firms, doles out $400,000 in fines, for AI washing, March 2024
- DLA Piper, SEC settlement highlights continued focus on AI washing, January 2025
- The D&O Diary, SEC files AI-washing enforcement action against restaurant technology company, January 2025
- Fortune, Albert Saniger, Nate shopping app fraud charges, April 2025
- Bloomberg Tax, App CEO lured $42 million from investors with bogus AI, SEC says, April 2025
- Akin Gump, SEC announces 2026 exam priorities, November 2025
- Sidley Austin, SEC sets 2027 exam priorities, October 2026
- McKinsey & Company, The signal in the sell-off, April 2026
- Bloomberg Government, AI bots auditioning for Wall Street trading are mostly losing, May 2026
MakeBanc is a technology platform that connects capital owners with independent strategy operators. MakeBanc does not provide investment advice, custody services or banking services, and does not use AI to select strategies, make allocation decisions or move assets. Nothing here is an offer or invitation to allocate to any strategy.