Welcome to another edition of C&F Talks. Today, I have with me Gabe Lakeman, who's a Partner at Latham & Watkins. Gabe's going to be speaking at our forthcoming summit on
Data, AI, and the Future of Financial Services, which is being held in London on the 9th of June. Gabe, welcome.

Great to be here, Maurice.

Let's turn to our first question.

What are the specific AI challenges and opportunities, as you see it, for regulated firms?

Yeah. I mean, I guess there's been a lot of interest in the opportunities of AI, sort of, generally lately. And I think, I mean, they're kind of obvious areas you can think of in terms of the kind of operational efficiencies that AI brings to provision of financial services and some of the kind of immediate sort of concrete applications, including things sort of like back office and compliance functionality.

But I think some of the kind of, more interesting use cases are really thinking around, you know, the extent to which AI can be deployed in sort of client-facing provision of financial services and bringing things like personal advice, you know, improving access and sort of helping growth in that sense. I think turning to the challenges, I think it's quite interesting. I mean, I guess the main challenges that regulated financial institutions have when deploying AI, were often brought about from thinking about how the sort of financial services regulatory framework applies to deploying AI.

And if you look at historically, certainly in the UK, you know, regulators have been pretty forward thinking about AI. Someone like the FCA has been looking at, so I guess going back to looking at things like machine learning, so, you know, almost a decade ago, and has had this kind of steady theme of focus on the area, you know, ahead of the latest round of interest that's been provoked by things like ChatGPT and those sort of generative AI models taking up a lot of attention.

And kind of working through those waves of regulatory attention, where the regulators seem to be getting to now is not really proposing to bring tech-specific or AI-specific sort of regulation onto the rulebook, but rather applying their existing regulatory framework to the deployment of AI, which sort of taking a step back makes quite a lot of sense, because something like AI, to the extent it's to automate existing processes or processes that would once have been performed by a human, often there's already existing rules to govern that activity.

And so those rules can be applied or read across to the AI use case. But the kind of rules you see getting read across are some of the ones that are some of the kind of broader or overarching aspects of the regulatory framework. So, things like the senior managers regime, like SMCR, which was the regime introduced to ensure the regulator could hold senior individuals accountable for firm actions from the financial crisis.

Or things like the consumer duty, again, which sets out very broad, overarching regulatory requirements that can be applied to almost any area of what a firm's doing, if it's dealing with retail clients, requiring it to ensure that it gives good outcomes for retail clients. So those are the kinds of broad regulatory requirements that I think is quite straightforward for a regulator to read across and apply in an AI context. So how do your senior managers ensure that they're discharging their responsibilities onto the senior managers regime as an organisation deploys AI?

Which on the one hand is, is great, because it's, you know, regulator firms are familiar with lots of these rules already. But the sort of challenge it gives you is, how do you tailor or work out how those rules, those very broad rules apply to new technology and new use cases that won't have necessarily been envisaged at the time the rules were drafted, in a context where you may have almost a kind of element of 2020 hindsight being applied if things, things do go wrong. And I think that's, that's really the challenge for the financial services sector.

They can kind of, to give an example, I think, think of something like, like robo advice, which I think is, it's quite interesting. That's an area where, you know, the regulators being focused on operational efficiencies and the provision of financial advice to investment advice to retail customers, that's serving a very clear, you know, public policy, good of broadening access to those services and those opportunities. And there's existing rules around that.

But those rules were developed in the context of thinking about advice being given by, you know, effectively deterministic decision tree that just goes in response to questions, right. And then if you bring along, you know, generative AI, that may be sort of not deterministic in the same way, it's not really clear how you apply those rules, how you do that testing, how you demonstrate that the outputs of a deployment of a model in that context, is really going to be meeting all your requirements around suitability, vulnerable customers, clients, best interests, consumer duty, and so on.

So, I think that's just a nice example of where you can see a very promising opportunity for deployment of AI. But the challenge really coming from how do you apply that existing regulatory framework to that?

Yeah, no, I think that that's a very good example. I hope, though, of course, that the regulators technology, if not, so that makes life a little bit easier.

And given that banks and many others are weighed down by their legacy systems, where are the traditional financial services firms in terms of their journey, in comparison to fintechs?

It's kind of interesting. I was thinking about this kind of topic just before the call, and I think there's a general trend in terms of kind of both traditional financial services firms and fintechs. Now, I guess the line started to blur a bit more as the services that they provide start to overlap more.

And as you get some of the more well established fintechs, they're now very large financial institutions in their own right. And so, I sort of think, if you're looking at those large financial institutions, it's a bit, it's more of a spectrum in terms of how far along people are looking at how they're deploying AI and how they're putting governance structures in place. I mean, the large traditional financial services firms have very well developed, you know, compliance frameworks and governance frameworks and senior manager frameworks.

And so, to some extent, adapting those to AI, I think, can be relatively straightforward to work through. And people advance doing that. I think, on the kind of more fintech side, and certainly as you move to some of the smaller firms into more startup space, that's where you can see quite ambitious plans to deploy AI, you know, often being led by the business.

And I think that's where you can get some challenges around how the kind of legal function, which may be quite small in the startup, really, really sort of gets its arms around, you know, what the organisation is doing and what the correct framework is to do that. As I said, I think it really is more of a spectrum.

Yeah, yeah.

And how do you think regulatory change might have an impact on the market in 2025? We see these sort of competing aspects, or maybe not competing, but we certainly see a focus on growth as being a very important objective for regulators. How do you see regulatory change unfolding over this year? And how might it affect this area?

Yeah, I mean, I think you've really put a finger on that. You know, how does the growth agenda feed through into regulation of AI? Looking back a few years, we had when the kind of FCA and the PRA and the Bank of England were putting out discussion papers on AI, they're really kind of thinking around the extent to which you needed sector-specific or tech-specific regulation, which obviously is not traditionally the way we do things in the UK framework and more broadly.

And it seems like, you know, the conversation has been consistent with that, in that people aren't proposing to put out specific regulation in the financial services sector. So then in terms of what does that mean for regulatory change? I mean, to some extent, if you'd asked me before the growth agenda, I'd be saying, well, it's not clear how much development there's really going to be in AI regulation over the next 12 months.

I think the growth agenda has now impacted that a bit. And obviously AI was mentioned in the FCA's sort of letter on how it's going to help achieve growth. And the FCA is now doing some work with the ICO in terms of looking at regulatory barriers to AI.

And one of the ones that they've identified on is that consumer duty requirement that I mentioned earlier. So definitely you can see the regulator appears to be open to how it can facilitate adoption. In terms of are we going to see concrete changes to the FCA handbook or anything coming out of that, there I'm less sure.

I mean, there's nothing in the FCA's regulatory initiatives grid, for example, that's come out recently that's really going to be focusing on that. So, I sort of think probably in 2025, I mean, I'm not expecting a huge amount of regulatory change, specifically around AI. I think it's probably going to be more a process of firms getting more comfortable that you can actually comply with the existing regulatory framework in AI, but you just need to kind of work out, work through the ways in which you do that.

Yeah, yeah.

And what about the biggest opportunities for innovation? I mean, you've mentioned some of them, you know, the personalisation of advice, wealth management advice and operational efficiencies. Is there a sort of standout area that you think is going to work well, you know, taking into account, of course, the regulatory implications of that area?

So, I think the kind of easy targets and the ones that people have been focusing on, just think of any area where you need to interrogate large data sets and the provision of financial services. And so, kind of clear examples of that are some more compliance of back-office type functions, thinking about fraud detection, you know, AML requirements, customer screening, transaction monitoring, push payment fraud, all those kinds of use cases. And the advantage of that is that there aren't really so much in the way of specialised regulatory requirements on how you deploy technology in those areas.

So, it's more, I think, straightforward for people to deploy. And people already, you know, have been doing this for a while now in terms of using machine learning techniques to interrogate large data sets to detect instances of fraud and market manipulation or whatever else it is. So, I think that's definitely a huge opportunity and one that's already being seized.

I think then kind of moving beyond that, the bit that I get quite excited about is how are you going to start deploying AI, again, in those client facing or customer facing roles. And you can see, so if it's going back to something like advice, you can see how, you know, there's an interim step where you still have to do the kind of human in the loop approach. So, you can have tools that are assisting a financial advisor, you know, or a trader or whoever else in performing their activity, where that is a data source using a lot among others. And again, people are starting to do that at the moment. And that's super interesting.

I think the kind of the next step is, well, how do you then put those, those tools, those models directly facing the customer. And again, people are doing that, you can think of it in terms of kind of chatbots and support functions but putting that straight into regulated services. I think that's the challenge. And, you know, what will be great, and we see people doing it.

There are also, as you can see, people thinking about innovative models, where if you can stay outside the regulatory perimeter, then, you know, can you start to do some of those client facing services in a way that doesn't introduce that level of overhead. So, investment advice around financial instruments, tricky, very regulated sector. But if you're thinking about budgeting advice, you know, that's not regulated in the same way, you can see how you could plug in inputs into using open banking to look at someone's transactions they've been doing and generate tailored advice on the base of that client facing, bit similar to what you'd be doing if you're trying to give people investment advice, but doing it in an area that's less heavily regulated. So again, I think that's interesting.

And all in ways which I guess will benefit the consumer, which is the overriding objectives.

So, Gabe, any further or final takeaways for financial services in 2025? What would be your headline?

I mean, I think certainly if you're talking to people from, I guess, legal and compliance background, I think now's the time, if you haven't already, to really get your ducks in a row in terms of putting in place those internal governance frameworks. Just some of it's quite simple stuff, like working out where exactly in your organisation AI is currently being deployed, because we see, you know, for a number of clients where actually start doing that, it turns out that people are using it in a large number of use cases, you know, way ahead of any kind of internal thinking about how to do that in a joined up manner. So, I think there's a fair amount in terms of getting ready to have those processes of governance and compliance structures in place.

And then beyond that, I think it's just going to be quite an interesting, exciting time as we see people start to really push out AI, get comfortable with it. We've had an initial wave of excitement around AI, and now we're kind of moving into a region where hopefully regulations will be relatively stable, and so there's a great opportunity for people to start really thinking through how they apply to AI and rolling out some of those services.

Yeah, so a year maybe of consolidation for people as they implement their AI strategies. Gabe, that's very, very interesting. For more on that and related issues, just for our viewers, please do consider attending the conference itself, which is the Data, AI and Future of Financial Services Summit being held in London on the 9th of June.

Further information is available from our website, www.cityandfinancial.com. Gabe, look forward to seeing you on the 9th of June. Thank you so much for joining us today.

Likewise. Thank you very much.