Waters Wrap: The rise of AI washing… and regulation washing?

Despite being a dirty, no-good Buckley/Reagan conservative, I consider myself to be an environmentalist. I’m militant about recycling and composting. I actively partake in and contribute to the reuse community. I shop local, never order food delivery, and—while definitely not a vegan—I monitor how much meat I eat in a week (though, that’s also thanks to my doctor). And when I get my daily iced coffee, I bring my own mug because single-use plastics are just the worst. 

I truly hope my smugness is coming through. The reason I consider myself holier than thou is because I live in Williamsburg, Brooklyn—or, as I like to call it, the land of hypocrisy. I have a lot of left-leaning friends and acquaintances. Actually, in Williamsburg, they’re all left-leaning, to say the least. And when in conversation, they talk a big game about the environment. The funny thing is, many of them don’t compost, don’t mind getting food delivery from places that still use plastics and Styrofoam, and order from Amazon like it’s a goddamn religious experience. Hipsters to local businesses: drop dead. 

Despite being my friends and neighbors, I believe these people partake in a micro form of greenwashing. Are they worse than a greenwashing, multi-billion-dollar corporation? Of course not. But when talking in groups, they try to gain clout by talking about how The Man has screwed us and the oceans are going to wash us away. (“Oh, hold on, I have another Amazon delivery at the door!”) Basically, they’d rather point a finger than do the difficult thing of walking the walk. 

Yes, I agree… I’m amazing and should be canonized after I’m dead.

But I’m not here to write about greenwashing—though that felt quite nice to get off my chest. Rather, I’d like to discuss the latest craze in false advertising—AI washing, which means making false and misleading statements about a firm’s artificial intelligence capabilities. 

This feels like the SEC is using a couple of small advisors to make a statement to the rest of the industry

Last week, the US Securities and Exchange Commission issued a scathing press release announcing that the regulator had fined two investment advisors a total of $400,000 in civil penalties. Delphia (USA) Inc., which is based out of Toronto, had to pay $225,000; San Francisco-based Global Predictions was ordered to fork over $175,000. The two firms “consented” to the fines “without admitting or denying the SEC’s findings”. 

In the release, SEC chair Gary Gensler said that the two firms made false claims to clients as to their AI-driven investment strategies. “Investment advisors should not mislead the public by saying they are using an AI model when they are not,” Gensler said. “Such AI washing hurts investors.”

In the same statement, Gurbir Grewal, director of the SEC’s Division of Enforcement, said the agency was “committed” to making sure that financial services firms were honest about their AI-investment capabilities, and took something of a victory lap. “As today’s enforcement actions make clear to the investment industry (emphasis my own)—if you claim to use AI in your investment processes, you need to ensure that your representations are not false or misleading.”

The SEC said that from 2019 to 2023, Delphia “made false and misleading statements in its SEC filings, in a press release, and on its website” that were forms of AI washing. As an example, the SEC said that Delphia claimed to use “collective data” to train its AI so as to make better investment predictions and provide more informed advice. “[T]hese statements were false and misleading because Delphia did not in fact have the AI and machine learning capabilities that it claimed.”

The release stated that Global Predictions falsely claimed to be “the first regulated AI financial advisor” and that it falsely claimed to use “[e]xpert AI-driven forecasts”. 

What are we talking about, here?

The Financial Times wrote this in its Alphaville commentary section: “The puniness of the fines and the triviality of the companies involved might make this easy to dismiss, but it’s the first such explicit AI-mismarketing charges that Alphaville can remember seeing from the SEC.”

Now, I’m a skeptic—and at times, an outright cynic—about most things. But, yes, the first thing that jumped out at me after reading the SEC’s release was “the puniness of the fines and the triviality of the companies”. Call me Captain Renault, but I’m shocked—shocked!!!—to find that financial services firms are not being completely truthful about their AI investment capabilities in here!

A few things, and these are simply my thoughts and opinions on the subject. So if I sound stupid or am confused on a matter, please let me know:

1. In all seriousness, but I believe most firms to be full of crap when it comes to their AI claims in press releases and on their website (though, to do it in an SEC filing is… special). Do you know how many press releases I’ve read just in the last month where some company is claiming to be the leading provider of… you name it? The reason we have our “This Week” round-up of company announcements is to strip out all the PR BS to just give you the actual salient points. Unless egregious, how do you realistically monitor something like this in the world of cutting-edge technology?
    
2. The SEC never explains how it knows that Delphia “did not in fact have the AI and machine learning capabilities that it claimed”. I do understand that that is likely due to the settlement agreement, but here’s a question: How do we know that the SEC has the sophistication to monitor capital markets firms’ use of AI? It’s not as if the top-tier ML engineers are lining up to work for a government job. And wouldn’t investors be better served if the SEC showed how these companies twist the truth about their AI skills, rather than simply saying that these companies aren’t fighting back on the allegations?
    
3. Is there a defined line as to what constitutes AI washing versus, I don’t know, AI fibbing? AI embellishing? Or is this Supreme Court Justice Potter Stewart saying that he wouldn’t define what entails hardcore pornography, but “I know it when I see it”? Yes, there are deep learning neural networks and generative AI, but there are also decision trees and simple linear regressions, much less basic robotic process automation—where is the line defining sophistication?
    
4. And to that point, how will vendors providing tools to investment firms have to prove their AI capabilities? Yes, there’s a certain level of oversight mandated before an algorithm is released into the market, but in the case of Delphia and Global Predictions, these were largely marketing and filing wrongdoings. Also, who gets fined when an end-user claims some sort of AI sophistication but relies on a third party embellishing its own wares? That usually still falls on the end-user, but shouldn’t tech companies need to be honest in their marketing, too?
    
5. Reading this press release, I couldn’t help but think about this article we published last October about Virtu Financial and its CEO, Doug Cifu, saying that it would take on the SEC directly because the firm vehemently disputed the SEC’s claims that employees had “unfettered” access to consumer data. I would love to see the SEC take aim at a company with deeper pockets on the subject of AI washing.

To be clear, I might be reading this whole thing wrong, but this feels like the SEC is using a couple of small advisors to make a statement to the rest of the industry after issuing an investor alert in January about AI washing. It’s a warning shot—which is actually what the FT called the announcement in its article. To me, though, a warning shot is nothing if firms keep doing this and a kill shot doesn’t follow. Yes, this could very well be the first of many fines, which could increase in amount and aggressiveness. Indeed, Inc. wrote this: “both firms now share the unenviable fate of birthing AI-washing, which will likely become a phrase cited as often as AI claims themselves.” (No chance on that latter part, but I get what they’re saying.)

Finally, I reached out to a handful of people I respect to get their read on this announcement, and several thought that I might be a bit in left field here. Fair enough… I enjoy being the holder of minority opinions. I didn’t include their quotes because this is an opinion piece, and I might be a bit too aggressive. But we’ll definitely be writing more on this subject. 

Ultimately, I guess that we’ll learn more about the SEC’s fortitude on this matter over the next 12 months—is this truly the first of many, or simply a message sent?

Have thoughts? Hit me up: anthony.malakian@infopro-digital.com.

The accompanying this column is “A courtroom scene” by José Guadalupe Posada, courtesy of The Met’s open-access program.