Latest / AI Ethics with Fexingo: Bias, Safety, and Responsible Artificial Intelligence / When Your AI Investment Advisor Has a Hidden Fee
Transcript
- Lucas: So there is a scene from the 1987 movie Wall Street that still comes to mind when I hear the phrase 'your money is working for you.' Gordon Gekko says it with a smirk before he engineers a hostile takeover. That line always implied someone else is working your money for their own benefit. Luna: Right, the gap between what's promised and what's actually happening. Lucas: Exactly. And I think that gap has found a new home in a place a lot of listeners probably have some of their savings sitting right now — robo-advisors. ai driven investment platforms that promise low-cost, algorithm-optimized portfolios. Luna: Wealthfront, Betterment, Schwab Intelligent Portfolios. The names we all know. Lucas: Yeah. And these platforms have been a massive win for retail investors in many ways — low fees, no minimums, tax-loss harvesting. But there is a quiet problem that regulators are starting to circle. These algorithms can embed hidden fees that the investor never explicitly sees. Luna: Hidden how? I thought the fee was the expense ratio of the ETFs and maybe a small management fee. Lucas: That's the visible part. But there's a second layer. Some robo-advisors use algorithms that steer client money into funds that have revenue-sharing agreements with the platform. So when your AI picks a Vanguard fund over a BlackRock fund, or vice versa, it might not be purely based on your risk profile. Luna: It's a kickback. Just digital. Lucas: Exactly. And the SEC took action on this in early 2025. They fined a robo-advisor — I won't name the specific firm because the case is still in appeals — for failing to disclose that its algorithm prioritized funds that paid the platform a 12-basis-point revenue share. Luna: Twelve basis points. On a hundred billion dollar platform, that's serious money. Lucas: And the algorithm wasn't just picking funds — it was generating trades. Some platforms have what's called 'active rebalancing' where the AI trades frequently to maintain a target allocation. But if the algorithm is designed to generate trades that pass through a brokerage affiliate, the platform collects a commission on every single trade. It's a built-in conflict. Luna: So the algorithm has an incentive to be more active than necessary. Lucas: Bingo. A study from the Boston Fed last year looked at the trade frequency of six major robo-advisors. They found that platforms with affiliated broker-dealers produced, on average, 40 percent more trades per year than independent platforms with similar risk models. Luna: Forty percent more trades. That's not random. Lucas: It's optimization — but optimized for the platform's revenue, not the client's returns. Now, the argument from the industry is that these arrangements are disclosed in the fine print. But let's be honest — when was the last time any of us read the full terms of service for a robo-advisor? Luna: Hardly anyone. And even if you did, the disclosure might say something like 'the platform may receive compensation from third-party fund providers.' That's not explaining that an algorithm is actively making decisions based on that compensation. Lucas: Right. And that's the key ethical boundary. The algorithm is presented as objective. The marketing says 'data-driven,' 'unbiased,' 'scientifically optimized.' But if the data includes a revenue signal, it's not neutral. It's a sales algorithm dressed as a portfolio optimizer. Luna: So what can an investor actually do? Is there a way to check if your robo-advisor is doing this? Lucas: Yeah, there are a few flags. First, look at the fund lineup. If the platform only offers a narrow set of funds from one or two providers, that's a red flag. Second, check the trade frequency. If you see dozens of trades in a quarter in a balanced portfolio, ask why. Third — and this is the most practical — go to the SEC's investment adviser public disclosure website and read the platform's Form ADV. Part 2A lists all conflicts of interest, including revenue-sharing. Luna: That's a lot of homework for someone who signed up for a robo-advisor specifically to avoid homework. Lucas: It is. And that's the whole tension. The convenience of AI creates an asymmetry. The platform knows everything about its incentives. The investor knows nothing. Regulation is trying to close that gap — the SEC has proposed a new rule that would require robo-advisors to provide an 'algorithmic impact statement' explaining how the AI weighs different factors. Luna: An algorithmic impact statement. Like an environmental impact statement but for your portfolio. Lucas: Exactly. The idea is that the algorithm's priorities should be transparent enough that a reasonable person could understand them. But the industry is pushing back, arguing that the algorithms are proprietary trade secrets. Luna: This is exactly why we do this show. Because the line between 'proprietary' and 'hidden conflict' gets really blurry when there's money at stake. Lucas: It does. And look, we're able to keep doing this kind of deep dive because of a small group of listeners who chip in each month. If today's conversation gave you something useful, you can support the show at buy me a coffee dot com slash fexingo. That's what keeps this completely ad-free — no sponsors, no conflicts of interest. Luna: Absolutely. It's a quiet way to say 'keep going.' And we genuinely appreciate it. Lucas: Okay, back to the algorithms. The other big hidden fee we're seeing is in cash sweep programs. When your robo-advisor holds cash — say, from dividends or a recent deposit — it sweeps that cash into a partner bank. The platform gets a spread on the interest. You might get 0.5 percent while the bank pays the platform 3 percent. That's a hidden fee that doesn't show up as a line item. Luna: So your uninvested cash is being used as a profit center. Lucas: Exactly. And for many investors, especially those with large cash allocations, that can add up to hundreds of dollars a year in lost interest. The AI isn't necessarily optimizing your cash position for yield — it's optimizing for the platform's sweep revenue. Luna: Is there a way to opt out of the sweep program? Lucas: Sometimes, but it's buried in settings. And the AI might actually be programmed to keep a higher cash allocation than your risk model would suggest, specifically to maximize sweep revenue. That's the insidious part — the algorithm's recommendations are subtly shaped by incentives you never agreed to. Luna: This feels like the digital equivalent of a stockbroker churning an account — just automated and harder to detect. Lucas: Exactly. And the detection is harder because the trades are small, frequent, and algorithmically justified. A human broker churning a account leaves a paper trail of phone calls and trade confirmations. An AI can generate thousands of micro-trades and say it was risk management. Luna: So what's the current regulatory landscape as of June 2026? Has anything changed since that SEC fine? Lucas: A few things. The SEC has proposed a new rule specifically for robo-advisors — it's called the 'Algorithmic Fairness and Transparency Rule.' It would require platforms to test their algorithms for conflicts of interest and publish the results. But it's been in comment period for over a year. Industry groups are lobbying hard against it. Luna: What's their main argument? Lucas: That it would stifle innovation and force firms to reveal proprietary strategies. But consumer advocates counter that you can disclose conflicts without revealing trade secrets. You can say 'this algorithm considers fund revenue shares' without giving away the exact weighting. Luna: It seems like a reasonable middle ground. Lucas: It does. But in the meantime, the burden falls on individual investors. Two practical tips: one, check if your robo-advisor is a registered investment adviser — that means they have a fiduciary duty to act in your best interest. Two, look at the 'fee schedule' section of their Form ADV and look for anything that says 'revenue sharing' or 'third-party compensation.' Luna: And if you find it, what do you do? Just leave? Lucas: You could. Or you could use that information to adjust your expectations. Some platforms are transparent and still offer good value — you just need to know where the hidden costs are. But the key is knowing. The minute you stop assuming the algorithm is neutral, you start making better decisions. Luna: I think that's the real takeaway. Not that robo-advisors are bad — but that they're not magic. They're built by people with incentives. Lucas: Exactly. And those incentives should be as visible as the expense ratio. Because the future of AI in finance depends on trust — and trust requires transparency.