Katherine Forrest: Yeah, it sounds really like directly relevant.
Brian Daly: It was funny in a way. I did entertainment M&A and worked with one partner who did esoteric securities law opinions. That partner wound up moving to New York and going in-house and took me with him, and that's how I became a funds lawyer. And I was in-house for a while. Then I went to Big Law, and I was a law firm partner for over a decade and watched the industry, particularly the private side of the industry, just grow and grow and institutionalize. And there was an election. There was a chair who came in who was aligned with where I was, and we hit it off, and he offered me this job, and I took it and didn't look back, and we sold the house and moved to DC. So now I'm probably 14 months into the job.
Katherine Forrest: Wow, that's a big move.
Brian Daly: It's a big move.
Katherine Forrest: Yeah, yeah. Let me take that now into the world of AI. Give us a sense of how AI is used in the Commission these days.
Brian Daly: So the Commission was set up in 1934 after the 1933 Act was passed. Obviously, a year later, and there's a '34 Act, and the Commission is—when you think of how the Commission operates and is structured, think of a company, a public company. And we have five commissioners, one of whom is designated as the chair. The chair you should think of as being the chairman of the board and the CEO, and the non-chair commissioners are akin to—doesn't track 100%, but in general, the other commissioners are like directors on the board of a big public company. They're important. They have to vote on everything we put out of any size. They have different areas they're interested in. Sometimes the chair will ask commissioners to go and run with a particular issue, but that's how we're structured at that level. That's the board. And then we have divisions and offices. We've got six divisions, three of which are policymaking divisions. So Corp Fin is the one most people know. Corporation Finance looks at the filings for IPOs, public companies, mergers and acquisitions. So most of what you see on the front page of the Wall Street Journal winds up rolling through Corp Fin, at least initially. Trading and Markets is exactly what you'd expect. That's the part of the SEC that works with the exchanges, works with the oversight—and it's the SROs that oversee the exchanges—and works a lot under the '34 Act for making sure that the market infrastructure and all the market participants are well regulated and heard. And our group is Investment Management. And so we are responsible for all the fund advisors and the funds. And that sounds like kind of a niche product, except until you run into the numbers. Very quickly, like, our market sectors—we've got 8.5 trillion dollars out in money market funds, which falls under our purview. Private funds, about 36 trillion. Registered funds—those are the retail funds—ETFs, mutual funds, other closed-end funds. And you overlay that with about 62 or so trillion of advisory relationships, and there's some double counting between that 62 and the rest of it. And we've got over 15,000 registered advisors we deal with. And in terms of regulatory assets under management, which is a post-leverage number, we're about 180 trillion with our market participants. So when people say the US is the capital markets leader of the world, that's what we're talking about. And so our division—I know this was a long lead-in—we oversee the rules and regulations around mutual funds, ETFs, closed-end funds, interval funds and tender offer funds, and business development companies. So that's the Commission and where we sit within it.
Katherine Forrest: That's huge. And how does AI then fit within that? It sounds like you could take AI and put it to work in any number of ways.
Brian Daly: So at the Commission, AI—it's actually operational today. It's not aspirational, it's operational. And it's interesting because I was there for some of the rollouts. Some of it predated me, but it was interesting how we handled it because it was very similar to what I've seen other companies do when they have this discussion about, we have AI coming in, it's a new tool. There's an acceptance that it has to be used in some way, shape, or form. And every company, every organization, I think, has a question of, well, how are we going to do this? Who's going to control it, etc. And at the SEC, it was the same kind of—your firm, you probably had the same discussion, your clients probably have the same discussion. And it's really a fundamental question of centralization and distribution. And I think we wound up in a good place because we have a split. We centralized things that should be centralized, like procurement, like cybersecurity, like information security, which is a subset of that, and some of the more impactful agency-wide impacts. Those are all centralized. And then after that it gets distributed out to the division directors. So I am personally responsible for the use of AI in the division. People misuse it, that's on me. I'm a supervisor, that's my job, and that's something I willingly accept. And if you can't accept that, you shouldn't be a supervisor. And that's the trade-off, because you don't get innovation, you don't get creativity when every use case has to be put into an application and sent off into some hole where somebody looks at it and who doesn't know your business at all, perhaps, and comes back with a parade of horribles over and over and over. It's important to be thoughtful and careful, and we are that way, but we do that in a slightly decentralized manner, and I think it's a pretty good balance.
Scott Caravello: So when we're talking about innovation, how do you see AI actually coming into play within your regulatory ambit? What are you hearing from the market? What are people telling you about how they want to be using AI and modernizing the system?
Brian Daly: We're pulling and we're pushing all at the same time, so we're trying to lead by example in some ways. We put out a lot of paper. We put out—now it's electrons. When we put out a rulemaking—so for us, a rulemaking, it's a big event. A rulemaking requires the Commission itself to vote on it. Rulemakings are governed by what's called the Administrative Procedure Act, which means we have to get notice and comment. That's a pretty good system in general. It's also the law, so we have to follow it. But it is a good system, and it means that we have to put together a proposal for a rule or a change or withdrawal, whatever we're doing. We have to support it in detail. Not only the merits in the securities law part of it, but we do an economic analysis, and we do other analysis that have crept in over time—Paperwork Reduction Act and a whole bunch of other things that are required. So we put out for a proposal something like 300 pages, and that goes out to the market. Historically, the way that always happened was it would go out, and a law firm like yourself would receive that. It comes out. We're trying not to put them out on Friday nights. Sometimes it just happens, but it comes out at 5:00. Invariably, even though we try and get the machine moving at 10 AM, it comes out. It hits everybody at the same time. It goes into the SEC email system today, and everybody gets it, and every lawyer gets it. In the old days, they would print them out, would just clip them, and they would take them home, and everybody would pull out the highlighter and start taking notes. And a few days later you'd have meetings inside a firm or some—or an advisor. People talk about it, and eventually client alerts would come out of it. Positions would come out of it. People would decide if they want to comment on it or not. We generally have comment—we always have comment periods for these kinds of things. Generally it's 60 days. I will say, invariably we get all the comment letters on day 59.
Katherine Forrest: I'm shocked. I'm shocked.
Brian Daly: That's how it goes out. That's a pretty good system, but that's the old way. So today, we look at a primary—and someday it will be the primary—consumer of our rulemaking, these 300-page tomes that go out, as being an LLM or some other AI agent. And we have actually—and you probably haven't noticed it, but you'll probably go look at it after we talk about this—we have been gradually changing how we present information in those to be more AI-friendly. So, one tiny little example is we have a date code now, and that sounds crazy, but for years and years, there was no date code. And if you had overlapping releases in different areas, you had to scroll through the whole thing. You had to find the signature block and look next to the signature block to see what the date was. Because if you had to chronologically order a couple of actions, that's how we had to do it. And the signature block is not necessarily the last page because there could be 100 pages of appendices. So it was frustrating, it was needless. And so now we have an actual date code which is recognized—like, the machines recognize it. So, in just a tiny little change, which I think will make life better. But we also have done textual changes. So if you go back, historically, SEC documents have been characterized by long substantive discussions in footnotes. We haven't taken those away completely, but we've reduced it by something like 95%. If it's substantive, it goes in the text, or we consider whether it needs to be in at all. And that's just to help the LLMs—and really human readers too—but to help people understand the argument, where it goes, what's important. So we've done that. We've got a lot more bullet points. We have shorter paragraphs, we have tables. Sometimes we'll put a table in the front, far more detailed than before, and we know it's working. So it's a sort of—in a way, it's structured data. I mean, it's like the lawyer textual version of structured data, but it's an attempt, it's a step in that direction. And we know it's working because now we do see that when we put something out, even if it is technical or complex or long, we see uptake quickly. It's on social media next day, same day. We see client alerts coming out from large law firms or accounting firms within days instead of weeks. And it's really good. I think that's good government.
Katherine Forrest: Do you assume that people are using another LLM to summarize? Is that really the—that's the basic—that's the basic point.
Brian Daly: People tell us that, yeah. And to switch to your questions, there's tremendous opportunities to make investing more understandable and to give retail investors more access, because retail investors buy products that are based on disclosure. We are a largely disclosure-based regime. That's how it's been since 1933, and it's worked really well for us. Obviously, we lead the world in capital markets. We do not have every IPO having to go to some state regulator who thinks about whether it can go public or not, which was the case prior to 1996. But those disclosures get longer and longer. They get more and more technical, and they're becoming electronic. And so the Commission put out for comment recently an e-delivery proposal. E-delivery is kind of what it sounds like. It's to move off a paper default and make electronic communication the default. You probably get it at your house from your investments or your plan. You get these envelopes that come in, they're generally blue plastic, and inside is a proxy statement.
Katherine Forrest: I read every word of them, every word. At the moment they come in, I open them up and I sit down. I put on a magnifying glass, and I read every word.
Brian Daly: I'm glad to hear that. You are the exception, but the default now—we, at least the proposal, would have it move to electronic. People can opt out. Older Americans, people who have whatever issue, they can opt out and still get it in paper. So that was important. We've been talking to the industry about this really since I came into the job. And we've said, e-delivery is great, going to save a lot of money. It's going to save a lot of postage. Documents in paper and plastic envelopes are not going to go into the landfill directly from the Postal Service without even being opened. Like, all good. But it would be a shame if that's where it ended. If we look back five years from now and what we've got is a system where people are just emailing PDFs, then we've missed an opportunity. Because—and I'm holding up my phone right now—this four-inch phone is the window for the retail investor. And a four-inch phone screen is not a good window to read something that is designed to look like an 8½-by-11, eight-, nine-, ten-point-font document. It just doesn't work. But what would work is some kind of AI agent. And we're encouraging people to come in with their ideas, because if my mother—you know, again, it’s her birthday, 86 years old—my mother's not going to read any of those documents that come in. Like, she's just not. But she might interact with whatever fund she has. That fund sponsor pops up on her screen and says, hey, Mrs. Daly, we have your annual report, and I can answer any questions you want. If you want, I can deliver you the full annual report. I can send you a PDF. We need some votes for some directors. But do you want to know anything? And things you might want to know—and you can ask me—are, how much money did I make? What were my expense loads? Who got paid out of my expenses? What conflicts do you have? How does this performance compare against other products just like it? And even the most tech-unsavvy person can do that, because I know every grandmother in America learned how to use Facebook to see pictures of their grandkids. So it's there. And we think that could be a tremendous advantage. And that can take this massive tome of very technical lawyer, you know, 40 Act lawyers writing for 40 Act lawyers' disclosure and translate it and make it accessible and get better understanding for the retail investor.
Scott Caravello: So people are coming to you to talk about that and coming in with their proposals. What are the questions that they're asking when it comes to the regulatory issues that that might present, whether it be marketing and liability? Where are their areas of concern?
Brian Daly: So the lawyer knee-jerk on everything is liability. Like, no matter what you say, the immediate synaptic response is liability. And I got that a lot when I started on this theme. But brain surgeons use AI. Brain surgeons can figure out the liability for AI. People who are running financial products can do the same. And the good thing about our system—it's been around for 86 years. It continues to work. We continue to see asset growth every year. Something is right. And I think what is right—one of the things that's right—is that we have a principles-based regime in essence, right. The core of everything our division touches is that the fund sponsor, the fund advisor, the board—like, whoever is involved who touches our division—has a fiduciary duty. So a duty to put the interests of the client first. They have a fiduciary duty, which has two pieces, a duty of care and a duty of loyalty. And so people come in and they say, what about liability? And we say, well, this is just another tool. People on this side of the table remember when lawyers couldn’t use email.
Katherine Forrest: He's pointing to me, Scott, not pointing to you.
Brian Daly: And we made it work. And why? Because it's principles-based, and we figured out how to use these tools. And so one of the goals we have going forward, when we write rules, when we give relief, is to be as forward-looking—and by forward-looking, it really means technology-platform-agnostic—as we can be, because putting out relief for today's solution, great, that's a six-month fix, but then we got to come back in six months. And when we do, it is probably already 30 days to 90 days out of date in the best-case scenario. So that's where we're going with that.
Scott Caravello: I can barely remember when we didn't use AI.
Katherine Forrest: OK, just barely out of the cradle here! You know, but it makes sense that the lawyers come in and are worried about liability, because what you don't want is—you don't want to substitute necessarily, at least at this point with the tools' development. You don't want to substitute, for instance, the entirety of a fulsome disclosure with just a summary. You still want the fulsome disclosure, right? And fulsome disclosure is the entire thing. And you can go to it and you can look at it and study it. So a lawyer, I think, would be worried about the substitution of the summary.
Brian Daly: No, of course. And clients, perhaps reasonably, perhaps not, when they are paying thousands of dollars an hour for a lawyer, they expect an answer from the lawyer that has absolute certainty behind it. If you do this, you will absolutely be OK. And the real world doesn't always work like that. So we do—we had a lot of meetings where people come in, and they've evolved very quickly, like everything else in this space. And in the beginning, people came in and they said, well, we need some help, we need some relief, we need a liability shield. And we had people come in. We took every meeting we could. Anybody who asked for a meeting, anybody has an asterisk—but it is rare that we turn down a meeting. Somebody has something to share with us, we can do it. We have Teams. We can Teams people. We can do a phone call. We will go to you. Sometimes they can come to us. We love getting smarter. We're smarter, we can do better regulation. So people would come in, and we'd say, OK, tell us—duty of care and duty of loyalty. You understand that. People coming in always do. Where does that fall apart, and why does this tool not work under that regime, when we need to get involved? And the message we gave back to everybody who came in was, come back to us with specifics. Tell us exactly where some—and it's never the big principles—but they'll say, oh, well, you have this guidance about this, and it's crosswise. OK, that's good. Tell us about that. We can fix that little thing. We'll make it go away. We'll make it more technology-agnostic or whatever. We can do that. But the big picture—there's no need for us to say AI falls within the scope, because it doesn't. It's just a tool. It's how you use it. If you're doing a good job and you're careful and thoughtful and you have good policies and you care about your investors, it doesn't matter what tool you use. The nice thing is, as we were having these discussions, people came in and registered. And so we have AI-native advisors out there now. And so, like, OK, well, there you go. People have done it. So they're out there. We have AI-native advisors. It's very exciting. We follow the news every day. We will often call people. We read the trades, we read the more, you know, mass media. And if we see something interesting, we'll generally call that advisor, and we'll say, hey, we see you're doing something—completely, you know, volunteer on your part—if you've learned anything, if you want to tell us about it, we'd love to have you come in or dial in. Tell us where you're going, where you see the market going, what challenges you've had.
Katherine Forrest: How does that go over, just sort of the random cold call from the SEC?
Brian Daly: Now, pretty good. The first couple… well, people freaked out a little bit.
Katherine Forrest: Haha!
Brian Daly: But we also picked people who we knew and we had—I had a relationship with professionally, but not so close that I was barred by my one-year ethics ban. So it's like people who fell in that sweet spot, and we put the word out. So after a month or two people got used to it. And this is voluntary, they don't have to take the call. I think everyone who has come in and spoken with us has left feeling really good about the division and how serious we are and how sophisticated we are. Because people will often come in with a goal of trying to convince us that the problem is real or the concern's real. And we usually will stop them after about a minute or two. Like, OK, let's fast-forward. What are the issues? Like, we're with you. Tell us what the issues are. What are your proposed solutions? What do you think? How could we screw this up? How could we make it better? And it's much more free-flowing, sometimes a little rambunctious, but I'm having fun, and I think we're doing good.
Scott Caravello: And so running through all of this is the Division's and the Commission's core focus on investor protection. So we've talked a lot about what the investment advisors are saying to you and what you're hearing from the industry. But when you think about the risks of AI reaching ordinary investors, what worries you? What is the division thinking about?
Brian Daly: So, some of the risks are the same as have existed since the first futures contract was traded in ancient Japan or wherever you trace it back to. But it's fraud, right? The smarter people get, the better tools they get, the easier it is to commit fraud, the easier it is to fool people. Well, that's just there. It's undeniable. That's a reality. That is simply what we have to live with in every—it's not just financial—in every sector. So we do worry about that. Of course, we have efforts ongoing in various divisions and offices on education. The private sector is very big on that. Clearly the industry cares about that. So that's important. Always have a worry that people start to put their brains on autopilot and let the tool do the work. I mean, my father used to tell me that when we were doing carpentry stuff, let the tool do the work. And he meant something different. It didn't mean turn your brain off and go away, because then you, like, cut your hands, which I have done. So that's a worry as well. I think the biggest fear that I have about AI right now is, frankly, non-adoption. I think there are a number of people who are just too concerned about the uncertainty of it. And it's funny, because these are people who price risk for a living, and they say, I don't want to move forward until I have an all-clear. And part of why I'm very happy that you called me and invited me on this podcast—part of what we're here to say is, please, go ahead, go push it forward. If you have a problem, call us. But we hope and expect that if there is a tool out there that can improve the outcome, particularly for a retail investor—pick up a few basis points or even more, reduce friction, get more people into the financial services sector, people, who would otherwise put their money just into a savings account, will actually invest—we want that to happen. We want to enable that.
Katherine Forrest: OK, so now I have—this leads me directly into the question that has been, I swear to you guys, the number one question in 2026. All right? And here I'm thinking of all of these complicated questions that we'd be getting in 2026. But the number one question from financial services companies has been about transcriptions. And they ask about, you know, what kind of risks are involved in using AI tools to transcribe basically everything, and what shouldn't they transcribe? And then there's the ultimate question, which is, is it a book and record? Does a transcription constitute a book and record? Because what do they do with it? And so, you know, I've got my response, but let me hear what your response is for the guy who really knows.
Brian Daly: So for those of you listening who are not in financial services—
Katherine Forrest: Good point, good point, good point!
Brian Daly: The SEC has had—we have multiples of these, but we have record-keeping obligations, record-keeping rules. And if you read the rules, the rules were written a long time ago. They were written pre-internet. And the record-keeping rules mandate what—in this case, let's say an investment advisor, somebody who advises you for a living or manages a fund—what records they have to keep. And when you read the record-keeping rules, it's actually shockingly short. It's a short number of categories. And the text is clearly written, like, expecting you to put carbon paper into the typewriter and, you know, save the extra copy of whatever you mail to Miss or Mr. So-and-So. It was a long history or development of this where there were questions as email came, you know, to light, people started using email in actual business, as to whether emails were books and records, whether they fell within the scope of those categories. And there was a long lead-up, and it ultimately ended, after a decade and a half of this, with some enforcement actions going out, mostly against broker-dealers, but a couple against investment advisors. And people are scarred by that. People have reflexive tics that come out when they think about record-keeping. So I'd say a couple of things. So number one is the chairman has instructed us to modernize the record-keeping rule, and any rulemaking that we have, we're not allowed to talk about till it comes out. And I would say that it would be strange to modernize the record-keeping rule and not think about commonplace uses of artificial intelligence and other products. Rulemaking doesn't happen overnight. Takes us a while to get our 300-page perfectus together.
Katherine Forrest: Do you use Claude to help you with that, or some other model? Gemini, you know?
Brian Daly: So, put a pin in that.
Katherine Forrest: Ah, OK.
Brian Daly: But no, people write these. But we have some interesting things about how we use it. But we also have the ability to give exemptive relief, and we can do that sometimes. And we do that where there's a problem in the market and we have authority to do so. We always have to think about whether we have authority, and the “we”—as the staff at that level, not the Commission. The problem with any relief that we give is it takes a good—by the time you get through thinking through everything and the internal sign-offs and being careful and thoughtful, like, you're talking a couple of months already. So when there's something on a rulemaking agenda, the problem has to be really bad for us to do exemptive relief, knowing that there is a potential rulemaking down the road. And the final piece of it is, I think when you look at the record-keeping rule, it's pretty clear about the categories that have to be retained. And I would just encourage people to look at those categories and think about, when they're recording, whether the recording falls in one of those categories. Because if it doesn't, it's not within the scope. People are going to be more conservative and save more things. We understand that, but that's up to internal policies. And what our exams colleagues do when they go out is they start with the policies that an advisor has. If the advisor's not following the policies—doesn't matter where the policy came from—that's a problem. So right off the bat, that's a problem. As a secondary level, they then go and look at the policies and say, is there a gap in these policies between what the law requires and what these policies say? And that's a second-order review. They always do it, and they always should, but they do that. I would say I've looked into this, I've asked, and I'm not aware of a single deficiency letter going out on this point. So I do think that people are reacting a bit to history. And our chairman has said multiple times—says it at least once a week—that the era of regulation by enforcement is over. People who are thoughtful and careful and who really put work, time, and effort into being discerning—to use that word again—on what their policies say and what they retain, they won't always be right. They'll be right an overwhelming amount of the time. When they're not, the way our system works generally is, when we see problems, we see industry practice moving in a way that we don't think is consistent with the rules, there's an escalation, a natural escalation. There are deficiency letters that come out from the exams team. When enough of those go out, they become convinced, or staff at large become convinced, that it's more widespread. Then public statements go out. Risk alerts and other statements go out to the industry saying, we know this is how people are looking at it. We don't think it's right. If you disagree with us, come in and talk to us. But if you think we're right, go—physician, heal thyself—go out and do that. And then only after that does the enforcement machine, in general, in most cases—does the enforcement machine ratchet up. Now, if there's fraud or anything else, you go right to enforcement. You know, we go and protect the retail investor. We always will. But that's generally how it goes. And so I would say, people who are standing on the sidelines with AI because of liability concerns, to really sit down with their advisors and think about—if there are some well-founded concerns, they should think about that. But if they're being overly conservative and they're losing out on market opportunities or they're losing out on the ability to improve retail investor outcomes, they should discuss that as well.
Scott Caravello: So one final question, I think, to bring all of this together, because you made a point in some of your prior speeches that you'd be disappointed if decades from now there's a future director who's giving the very same speeches that you have about AI and innovation and the need to keep up with technology. So bring it all together. What does getting it right look like so that future director isn't giving the same speech and delivering the same message that you are now?
Brian Daly: The first thing we want to do with anything, any initiative we take—do no harm, follow the code. We don't want to impede anybody. We don't want to cause any problems in the industry. So we think long and hard before we act. So if somebody looked back 10, 20 years from now and said, boy, Brian's initiative on X really screwed this up—that would be really sad and disappointing. And we hope that we don't do—we're very much hoping we don't go there. Beyond that, even if we get things right in the short term, but it turns out that the next director, two directors down, that staff has to come back and rework something because, in spite of the fact that we tried to be technology-neutral, we missed the boat, and we are taking time away from some other valuable rulemaking 10 years from now to come back and revisit this one, whatever we put out—that would be disappointing as well, because we're trying to put out durable rulemaking and take durable actions. And the durability really, you know, it's the seed that grows in the soil of common sense.
Katherine Forrest: Oh, I love that. I love that line. Durability is the seed that grows in the soil of common sense. That's good. On that note, I want to thank you, Brian, for coming in.
Brian Daly: Well, thanks for having me.
Katherine Forrest: Yeah, this has been incredibly interesting. And we're going to make sure that not only our audience, but if there's anybody who misses this, we're going to circulate it to our clients, because they're going to be really interested in all that you've had to say. So thanks very much.
Scott Caravello: Thank you. This was fantastic, and happy birthday, Mrs. Daly.
Katherine Forrest: Yes, happy birthday, Mrs. Daly. And just to note that your son would not sing for you on the air. OK. All right. Katherine Forrest signing off.
Brian Daly: Brian Daly as the guest. Thank you.
Scott Caravello: I'm Scott Caravello. Thanks for joining us.
