Ep 286- FDATA Executive Director Steve Boms

**John** (0:00)
Hey, it's the FinTech Newscast. My name is John, and with me as always is Mr. Perez. Welcome to the podcast.

**Steve Perez** (0:07)
That is Steve to you. Thank you very much. Good to be here.

**John** (0:11)
Yes, yes. Well, I'm going to refer to you as Mr. Perez. Just for this episode, because of our very special guest, we're lucky to have this week, the Steve Boms, the executive director at the Financial Data and Technology Association. Welcome to the podcast.

**Steve Boms** (0:29)
Very happy to be here, and I can be Steve B and Steve Perez can be Steve P, although I don't know that that makes it a lot easier.

**Steve Perez** (0:35)
No, it's actually more difficult.

**John** (0:36)
I thought you were going to go with A and B.
Okay. Well, what we've seen in the news lately, just to start off with, is agentic trading, some announcements being made. And this is kind of surprising. I guess I knew this was going to happen, but Robinhood announced that they would use agentic AI or make that available for trading on a certain account. You can set aside a specific amount of money, which you better do, in this case, for the agent to do its thing. Would you trust anything like that, Mr. Perez?

**Steve Perez** (1:14)
I think I would, especially if I gave it very clear instructions and a small part of money to work with. It probably wouldn't work with my entire portfolio, but I actually do like the idea a lot.

**John** (1:25)
Really? Why do you like that? It seems kind of scary to me.

**Steve Perez** (1:29)
It seems scary, but I think we were discussing before we began recording that there's sort of a massive distance between rules-based trading and just agentic trading with rules-based trading, right? You say, buy at this level, sell at that level, whatever. We have all these triggers. Once you have an event that actually falls outside that, the rules you've created, the whole thing sort of breaks and stops. So I like the idea of using agentic in a way that will understand, you know, my trading patterns, my goals, et cetera, and make decisions based on that without a whole lot of guidance from my side.
And also looking at, you know, things like unstructured data as well. So I'm all in. Sign me up.

**John** (2:07)
More faster insider trading. I see your point there.

**Steve Perez** (2:10)
More faster insider trading. Yeah. I do wonder though.

**John** (2:13)
Mosaic theory, yeah.

**Steve Perez** (2:16)
Which is what?

**John** (2:17)
Oh, you can put together pieces of public information to have trading positions that look like they're based on non-public information, but you put together so much public information that that's legal because that's just your efforts in putting together data.
So yeah, what you're saying about unstructured data so we could go out there and pull information. Kind of put together that something non-public is going to happen based on the public data that maybe an AI can detect that. And so we have to re-think about what's public and what's non-public information at that point.

**Steve Boms** (2:56)
Although, John, if AI can detect it and now AI and agentic trading is available to all of us, is that not public information at that point?

**John** (3:06)
Yes, that's the defense. It's called mosaic theory, that you just put together all the pieces or your AI did. And so the AI should go to jail, your honor, not me.
It was the AI that did it.

**Steve Perez** (3:22)
Are these agents actually liable if they do something that's illegal? Like if they actually say that you work for a company, or they sort of use your training or trade in a way that would be considered illegal or unethical, does that mean that they're liable or are you liable? It's sort of like the driverless car thing. Like who's liable for that?

**Steve Boms** (3:41)
I love this, Steve Peek. We're spending so much time thinking about this at Fdata. So I guess the way that we think about it is the concept of agents as a general matter in financial services are not new. They've been around forever. It's just this is a much different context. They're computerized, they proliferate, they think and act more quickly than we do, and there are fears that will go beyond the bounds of what you've instructed them. But there's actually a really fulsome set of liability, fiduciary responsibilities and responsibility for agents in financial services law and regulations. And so a lot of the work we're doing right now is like tracing, are there any gaps in the agentic commerce age that don't fit based on where agent law has been forever?