Get the Monday email →
Loading player…

Can AI really disrupt financial services? presented by WP Intelligence | DC Fintech Week 2026

DC Fintech WeekFri, Nov 6, 2026, 12:00 AM

The potential is there. Meta’s new AI agent, Muse, for example, could change how people shop, pay and choose financial products online. But banks and payment systems are hard to change: people trust what they know, regulations are strict and the biggest players are deeply embedded. We’ll look at what it will take for AI to break through those barriers. • Dan Gallagher, Lead AI & Tech Analyst, WP Intelligence • Benjamin Guggenhein, AI & Tech Brief Author, WP Intelligence • Luiza Savage, Editorial Director, WP Intelligence • Samantha Pelosi, General Counsel and Chief Compliance Officer, Navro

Transcript

Good morning, everybody. I'm Louisa Savage. I'm editorial director of Washington Post Intelligence, which is the Washington Post's new analysis and briefing service for decision-makers in business and policy. We offer in-depth intelligence reports with actionable recommendations, weekly briefings for our subscribers. We are also home to the WP Intelligence Councils, which is a membership community for executives.

We come together on a regular basis to hash out issues facing decision-makers across our coverage areas, which are AI and technology, energy and climate, global security, and also health care, which are increasingly becoming one story. One big story that we've been following and we're very interested in unpacking here at at the at the summit is what will it take for AI to live up to its transformational promise across various industries. So, we're very pleased to be a media partner in DC Fintech Week, and we're here to explore with all of you the opportunities and the obstacles for AI in financial services. So, to hash out this question of, you know, what will it really take for AI to to transform financial services, we have a great expert panel here today. Samantha Pelosi is chief compliance officer at Navro.

It's a cross-border payments company. She has extensive experience in financial services, particularly in blockchain payments and anti-money laundering. And previously, she has held leadership roles at the Federal Reserve Board, the US Treasury, and the American Bankers Association, where they developed innovative compliance strategies and they influenced global policies. So, that's really cool. We're really excited to have you here, Samantha, and all your breadth of knowledge and and experience.

Um and from our team, Dan Gallagher is our lead AI and tech analyst. He's based in San Francisco, just got in late last night. You may know Dan from his 13 years covering tech and media for the Wall Street Journal's Heard on the Street column. And he's since joined our team and we're really thrilled to have him. And next to me is Benjamin Guggenheim.

He is the author of the um AI and tech brief, which is our daily AI newsletter. It's a free newsletter that comes out every afternoon. And it's all about the intersection of AI, technology, and policy. Benjamin is a former congressional reporter based here in Washington, D.C.

and the recipient of the National Press Foundation's Everett McKinley Dirksen Award for distinguished reporting of Congress, which is all to say he's the most wired person I know when it comes to uh politics and policy machinations around AI and tech. So please uh put Benjamin on your speed dial, send him all your scoops, and and subscribe to uh his newsletter. Um so the first topic we wanted to explore is um you know, where the benefits and the value are of AI in financial services. And I wanted to start with you, Samantha. You've wide experience in the space and when you look across the financial services sector, which is obviously very big, where do you see AI delivering measurable measurable returns already today?

Yeah, thanks for the question. Um I because of my experience, I worked with a number of fintech and large banks on their AI algorithms for detecting um and for detecting money laundering, terrorist financing, and fraud. Um and this was 10 years ago, actually. Uh because it's Compliance tends to be a huge cost center for firms um globally, especially because there's reputational risk, there are huge fines that keep coming down. I I from the banking regulators, from Treasury, from FinCEN, which fights money laundering.

Um and so there've been a number of a number of attack vectors essentially that the that fintech and compliance tech have been attempting to cut down on. So AI is great at seeing patterns. So a lot of large banks were looking at besides the patterns you see with money laundering like money mules or um human trafficking, it looks the same like the money movements look very similar. Um what besides getting cutting down on false positives, how does um asking the machine, do you see any anomalies like that we don't know about? That's pretty cutting edge.

Um it tells you what you don't know. There's also a lot of great work with just how uh hackers and and that thieves uh take your credit card and put it you know, type it in your information and they've bought things off of the black web um and steal your financial information and they're cutting and pasting cutting and pasting or they have actually AI doing that for them. And there are other counterbalances to catch that kind of behavior that it's not a consumer typing in their credit card, it's a fraudster that that stole your money and they shut it down. So there's a lot of different this is all very back office and it's not really seen by the consumer, but it really helps bring down the cost of transactions.

That's fascinating. Um can you tell us you said this is a major cost center for firms. Can you tell us like the scale of that?

Yeah, I mean there it's regularly um at large financial institutions and large fintech, there will be hundreds of people compliance staff. Um and then there's all the folks that like our colleagues here that interface regularly with regulators and try to see like where the puck is going and it's just the this is the full-time job of the criminals and so they're always ahead of us and we always not just have to be on the defensive but have to be scanning the landscape and the fines are astronomical. I mean they just get bigger every year. They're now into the billions. When I was at FinCEN it was like 25 million was a big deal here in Washington D.

C. with Riggs. That's like nothing. That's like change.

Wow. Wow.

yeah.

And and so

[laughter]

you also said one other interesting thing before I moved on to Dan. You said that it's really helping the defenses by this ability to find these very subtle patterns and and and find, you know, whether it's money laundering or other criminal finance. Um but on the other side you've got criminals using AI.

Yes.

Um so is it really giving an advantage to the defenders or is it just kind of like escalating the the warfare?

I think it's escalating the warfare. I mean everyone is adapting to use AI, the bad guys and the good guys. But I think it's going to help pull down the costs of compliance and actually sharpen the focus instead of having all these false positives that have to have human eyes on them. The machine will be like, no, just don't even look at this one. But oh, we found something that you hadn't even thought of over here.

This is just a really strange pattern and we never see this. You might want to look at this. So it's it's very helpful.

I mean that's like doesn't it remind you of of like what we're hearing about radiology? Like finding tumors?

Yeah. That's like one of the very first AI applications.

Yeah.

cancer.

Yeah. Yeah.

[laughter]

So Dan, on this question of value, um you're in Silicon Valley. Where where is Silicon Valley and and where are Wall Street placing their bets on where AI will deliver value in financial services?

Well, right now I think they're placing bets all over the place because they're still I mean we're still in very early phase of AI. When you think about, you know, the models are still adapting, you know, the basic models underpinning and what kind of models companies are using. What I'm seeing and especially with Fintech is that there's two different broad areas. There's a lot of investment money going into disruptors. Um companies that would theoretically disrupt major players in existing players.

There's also a lot of investment going into companies that wouldn't necessarily disrupt but work into the system like being overlay. Um and I wouldn't know I wouldn't be able to say like there's not necessarily more money going one to one than the other. But we are seeing there's interesting investments into AI native um platforms that help with like insurance underwriting, financial services, um the the guy that backed originally backed PeopleSoft and Workday. These are two like kind of well-established Silicon Valley software companies has backed this company called Ridgeline which has raised some significant amount of money is now valued over a billion dollars. Um and then you've got if you listed like your top private top most valued private companies now you're still seeing companies like Stripe and Ramp.

These are very established Fintech players. Um I think Stripe is now valued around 160 billion in the private market. So these are there's a lot of action going on within that. Silicon Valley is really fixated on agents. I know we'll talk a little bit more about agenda commerce in a minute but AI agents that can actually do things for consumers, for business users.

That's where a lot of the direction is going right now for like in Fintech and and beyond.

Mhm. Mhm. Um so we've talked about a few of sort of like the visions and and where the value is is trending. Let's talk about some of the obstacles and and some of the concerns. And Benjamin, you keep close tabs on what lawmakers are up to and and we've seen a a very hands-off White House when it comes to AI regulation, famously.

Um, but the same can't be said for Congress. Um, what are lawmakers worried about in the fintech space, um, particularly with AI?

Yeah, so I think maybe it's helpful to touch a little bit on where the administration has been. In the spring and the summer, we obviously had a very kind of venture capital David approach to AI policy led by David Sacks and folks in the administration at the White House, which is was very anti-regulation. We want a light touch. We don't want to place any hurdles in the way of in innovation. But now we're kind of seeing that shift in real time.

And so I I just wanted to touch on this first before going to Congress. And that's since that we now have the super intelligence task force led by DNI director J. Clayton. Um, you know, it's co-chaired by the FTC commissioner. And uh, you know, Scott Bessent is in the mix.

So what that really tells me is that you know, we had this period in the spring and last year where it was it was a very hands-off approach. And now as the risks are getting real, the hugging face attack, uh, the administration and industry worried about some sort of financial Pearl Harbor from cyber attacks, the focus is shifting to a national security and uh, you know, maintaining the fine stability of our financial system lens. So that is becoming more important. On the lawmaker side, you know, Congress is very slow to act. There's a lot of polarization on AI policy in general.

But uh, you know, you do have some lawmakers who are really concerned about how AI could contribute to faster and unpredictable bank runs. So when you have thousands of agents potentially empowered to make trades and investments that are reacting to the same piece of news or and are modeled in similar ways, you could have extremely fast withdrawals from banks that could suddenly, you know, pose major problems for their liquidity. And so, uh Rep. Bill Foster and and particular member of the Financial Services Committee, Democrat from Illinois, has raised this issue of, you know, during the Silicon Valley Bank collapse, yes, they didn't manage liquidity correctly and hedge against risk, but a lot of the problems was, you know, portfolio companies and VCs all talking to each other and initiating these very quick digital withdrawals of their money extremely fast. I think it was 40 billion on the first day and then back orders of 100 billion.

So, that could be exponentially made worse by agentic bankruns. And so, what you're seeing is lawmakers raising alarm about, okay, we already needed after the SVB collapse better monitoring and embedded systems to determine liquidity. And with agents getting along coming along, it's going to get 10 times worse and a bankrun could happen in in a matter of seconds or minutes rather than days.

Yeah, I know that that's and and there is legislation that's been introduced and and I think passed through the house, right?

Yeah, the FUTURE Act would um kind of collect more information from Treasury and other financial regulators about how our system works. I think that's a kind of first step to collect information. That there still hasn't really lawmakers haven't really coalesced around a proposal to to deal with agentic bankruns. They're kind of more in the information collecting phase.

Interesting. Okay, we'll watch that space. Um Samantha, from your perspective looking across the sector, what are the biggest sort of hurdles to realizing the promise of AI.

Yeah, um there are quite a few, I think. Uh Just thinking through how even faster payments, when I was at the Fed, I was in the payments division and consumer protection, so I'll kind of bring both of that that perspective to this answer. Um with the payments with faster payments where smaller retail payments can settle instantly along with wires that settle like go out and settle instantly. Um With faster payments, now if you go to your Zelle app and it's hooked up to the faster payment system offered by the Fed and by the clearinghouse, um anytime you wanted just send a friend $5 or whatever

[laughter]

for your movie ticket, it's like asked you five times whether are you sure? Are you you're not sending this to somebody that has your grandchild in like a jail cell, right? Like it keeps asking you and asking you and asking you. And so banks and other fintech and the card companies are all trying to kind of protect the consumer, but it sometimes it becomes paternalistic and I think that will really slow down faster payments. It'll It'll slow down these agents.

I think another thing that is you're going to have to contend with almost immediately is in law, if you have if somebody is your agent, you are liable for whatever your agent does in, you know, business or what what have you. It's a like a very basic principle and so if consumers are going to be liable for what their agents do and we know it's going to go wrong, right? The Visa and MasterCard and the big card companies are all going to have to sit down and look at their like their chargeback rules and how are they going to deal with consumers that were like, well, I didn't really want to buy that stock or that you know, this and that and they have a lot of protections. Consumers have a lot of protections on pulling back money from their debit card, from a credit card. I mean, there are different rules, but there's a lot of wiggle room there and you have like sometimes, you know, 30 to 60 days to do it, to see see something on a statement, a paper statement and and raise that flag.

So, there's going to there's a potential for faster fraud also. Um there there are a lot of there's a lot of stickiness and and little complications in our payment systems, depending on what payments you're executing. Got it. Well, that's a great segue to you, Dan. You recently wrote a report for us at WP Intelligence entitled, AI Wants to Take Over Commerce, Humans Have Trust Issues.

Walk us through what some of the things that you were talking about, some of the obstacles that companies that are putting big bets on our agentic commerce are are faced with.

Well, I mean, trust trust is obviously a big thing and what's interesting is the trust has to be on both sides and I think that's one element sometimes missing from the discussion because um you know, I based this, you know, Metal launched its news agent, got a lot of got a lot of interest in the lot in the last month. There's been um it's one of the most downloaded apps people have ever seen. So, there seems interest in users, consumers making use of an agent somehow, whether it's just going to help comparison shop. Um there's an interesting safeguard in there. Consumers still have to approve any actual transaction, like on a final on a final click.

So, right now there is these safeguards about like, you know, an agent's not going to just go and empty your bank account to buy, you know, buy something. Um and there's you know, there's ways to put in like tokenized payments and everything to so you can put some limits in that. But, even if you get to a point where consumers have enough trust to like give some financial decision-making over to an agent. The whoever's on the other side of that also has to have the trust and be willing to deal with an agent. If you're talking if you're an e-commerce platform, if you are in financial services or travel, I think one of the biggest early use cases we've seen in the Gen Z commerce is people comparison shopping for travel.

Hey, find me the best the cheapest way for me to get from San Francisco to DC on this date. That's going to presume that whatever merchants, airlines, travel sites the agents crawling are going to be willing to do that so that it makes the you know, the more friction you have in a process the less adoption you have. So right now the the focus is going to be to get rid of as much friction as possible, but how to do that in a way that still safeguards consumers and still gives merchants the data they want is really the big question right now.

And another little detail in your report that I found kind of amusing was this idea that oh yeah, people actually really like to shop.

They do and I think um in fact the CEO of Affirm, a company in a fintech company, made this point at an investment conference where he said I don't you know, I'm fine letting an agent like go buy go buy milk for me. I I hate doing that. But when it comes to like me finding out the next bike I want to have if he's avid cyclist, I want that process. I want to be able to take the time to look at all that. So that again it's presuming that we want to how much do we want to give over to an AI agent?

It's not going to be universal amongst all users.

Sure, sure. Well, that's um a lot to unpack there, but just going back to the issue underlying both of your comments which is which is the trust on both sides, the payer and and the buyer and the seller. Benjamin, just going back to this trust issue, you mentioned how the administration is right now kind of starting to take some of the security issues more seriously around AI and and setting up this task force. Is your sense that regulators and policy makers are going to tackle this the trustworthiness of AI or do they think that they have enough authorities, enough regulations in place and that it's really up to the private sector to provide trust to consumers if they want their products to be adopted?

Yeah, again, I think this is very much the stance of the administration with the regard to this is very much changing in real time. So, you saw probably many of you read in the news there was a big day where all the tech company CEOs descended on DC and had this lunch at the White House and you know, Nvidia CEO Jensen Huang was there, Dario Amodei was there, Greg Brockman of OpenAI and they came out of this meeting with this White House Accord on super intelligence that basically said it's our responsibility to create this technology safely and steward it responsibly and deploy it responsibly. And so, yeah, that was very much a continuation of the administration's approach of listen, these guys are calling for pacing frontier AI development, calling for this regulation and we're skeptical of the reasons why they're calling for this regulation. So, we're going to we're saying no, it's your responsibility and if you don't follow up on your responsibilities, we can come up we can come after you with consumer safety laws through the FTC or come after you with the DOJ. Um I think that's already shifting because just in the past few weeks we've seen news of agents hacking government sites, both US sites and OpenAI agents I believe hacked into health insurance site owned by the Australian government.

And so, recently and this was an Axios report, uh the White House said no, actually notification of these incidents is mandatory. It's not voluntary anymore. And so I think we're getting to a point where, you know, these incidents get realer and realer and the administration, you know, doesn't want to have a hands-on approach, but it may start to have to have a hands-on approach if these incidents continue.

Mhm. Well, Benjamin, I just want to very quickly um ask you a look ahead question then we'll take uh maybe one question from the audience. So, please prepare your questions. Um midterm elections coming up in a few weeks. We don't know what's going to happen.

We're not here to predict, but uh polls suggest that Democrats could take the house. Um what does that change for AI and and fintech?

Yeah, so I think first of all, uh the oversight power that House Dems will have, they will subpoena They have already said they will subpoena AI executives to come and testify. So, that's one thing that's going to change. They're going to get to the bottom of these rogue incidents. Um you know, in terms of fintech and AI legislation in general, I I have a hard time kind of seeing any kind of AI legislation, fintech-related or not, moving through this Congress, but again, I think you know, Congress is slow to act. There's a lot of polarization on this issue.

Lawmakers, Republicans and Democrats are splintered within each party on how to handle this, but again, I think it's a matter of the incidents continuing. It's It's prompted change in the administration's stance before, and if they get really bad, uh Congress might act.

All right. Anyone have a question that that No? Okay. Um well, I'll close um with you um Samantha and Dan. Um as we look ahead to um you know, the value that we were talking about, the opportunities here, both business to save money and make new money, um who's going to capture that value?

Do you think it's um the big financial institutions? Is it specialized fintech companies? Um Or is it the large technology platforms and and AI companies themselves?

Oh. Thank you. I was looking at let you go first.

[laughter]

I think that's up for grabs actually. The I think right now the AI companies are not pricing like use of our agent. It's you know, tokens and time using agents for any kind of thing. And I have to say fintech and the banks, this is what they do, right? How do they price their services down to a fraction of a penny for a payment.

I I think that they will come out on top at least with the monetization right away just because they have the scale, the volume of transactions. They're very good at pricing their services and and moving the levers to extract the value. And again, I think it will make a lot of these large financial institutions that are kind of tech forward able to streamline their operations and bring down costs generally. Not by necessarily firing all of the workers, but just really helping with operations and and using their their human intelligence to a greater level than pushing papers and looking at tiny minutia. So I think it'll um it will pretty much at the beginning be the fintech and the and the financial institutions.

So Dan, we had one vote for the incumbents.

[laughter]

Incumbents or disruptors? We have 46 seconds on the timer.

I think mostly incumbents. And the thing I'll add is that as far as the technology platforms go, I think they for the you know, the Microsofts, the Googles, the Metas of the world, these things are going to be added to their experience as a way to keep users sticky in their ecosystem as a as opposed to like I think becoming a major profit driver. I do think we're going to see after we see especially after Anthropic and OpenAI actually go public and they will be going public. Um I think right now it'll be interesting to see the trend line in the pricing especially generic pricing because right now I think we're enjoying a bit of a subsidy because they have an interest in having big user numbers to put in their IPO prospectus and and at some point they're going to have to maximize the value that they create. So I think post IPO you're going to start seeing the prices go up as they try to extract more value out from what they can.

Excellent. Well, thank you all. I think we covered a lot in a very brief period of time. Um thank you all for for joining us. Uh please check out the AI and tech newsletter that Benjamin writes every day.

It's free. Check out it check it out on the WashPost website and you can find us at Washington Post Intelligence. So thank you so much for for having us. It was a pleasure.

Automatic transcript — names and jargon may be misspelled.