In this episode, EY UK Digital Assurance Leader Richard Harrison joins us to talk about how the firm is using AI agents in audit. How do they divide tasks between people and tech? And how are they understanding and mitigating risk?
Then, Polly Tsang, ICAEW’s Senior Financial Services Regulatory Manager tells us about Prediction Markets. While you may not have heard of them, they’re opening up a whole new market for Insider Traders. Polly explains how they work, where the risks lie and what needs to happen to minimise them.
Host
Philippa Lamb
Guests
- Richard Harrison, UK Digital Assurance Leader, EY
- Polly Tsang, ICAEW’s Senior Financial Services Regulatory Manager
Producer
Natalie Chisholm
Series Lead
Mark Rowland
Transcript:
Philippa Lamb: Welcome back. Today we're talking to EY about how to use AI agents in audit. How do they divide tasks between people and tech, and how are they understanding and mitigating risk? Richard Harrison, EY’s UK Digital Assurance leader, is with me. After that, Prediction Markets—never heard of them? Well, maybe it's time you did, because they're opening up a whole new market for insider traders. Polly Tsang, ICAEW's senior Financial Services Regulatory Manager is here to tell us how they work, where the risks lie, and what needs to happen to minimise them. Welcome, Richard.
Richard Harrison: Thanks for having me.
PL: Thanks for coming in. Now EY is using AI agents, as I say, it's a big change. Before we get into the how, can you just remind us exactly what AG Agentic AI is?
RH: I think a lot of people, when they hear AI, the first thing they think of is these chat interfaces where you can ask it a question, it comes back to an answer. So as we move towards agentic AI, rather than just answer a question, it can actually take an action on your behalf. So it has agency. So that could be as simple as drafting you an email, helping organise your calendar or something more sophisticated, like perhaps execute an audit procedure.
PL: This is a global rollout, you're not dipping your toe in— it's worldwide?
RH: That is correct. So we are deploying AI agents within our core audit platform, which is called EY Canvas. It's a cloud-based platform. We deliver about 160,000 audits a year out of that platform, and we've started deploying agents specifically into that platform, to everyone worldwide all at once.
PL: It's a big, bold decision, obviously with a range of drivers behind it, but can you just outline for us what the main business case was for you?
RH: We, at EY, have an absolute imperative to maintain, if not consistently improve audit quality. If you consider the world which we are currently operating within, companies are getting increasingly complex. They're going cross border, they're producing enormous amounts of data, and the level of order effort is going up all the time. So what we're trying to do, what we need to do is use this technology to really help us innovate and create capacity to help us keep up with those growing demands. Then just outside of the core audit work we're doing at the moment, we know this technology is materially impacting the companies we work with. So making sure we're absolutely at the cutting edge of this allows us to give sharp insights and perspective on how to do this responsibly and how to do it safely.
PL: Now obviously when you say you are using AI in audit, you are not using AI for all audit tasks. What are you using it for? What sort of things?
RH: We've done this iteratively. So the first agents we deployed were helping us do things like help our auditors go research topics. So accounting standards, audit methodology. We then started to move towards using it for project management type activities. So inviting people into the file allocating work, sending or drafting requests that we send off to the companies we audit. Now we've started moving to the next phase where we're starting using it to support us preparing audit work, but also reviewing audit work or helping us with our self review.
PL: Okay, so how does that work in practice?
RH: Yeah, let me bring that to life. So if you're an audit professional and you're preparing your audit work, traditionally, and if I look back to when I was a graduate, you'd finish your audit work and you hand it off to your manager to review. And it'd be a bit like a school kid handing in their homework for a teacher to mark, right? Then you'd wait for the manager to come back with comments or things you might have missed, not thought about. Then you'd go back to the company you're working with and ask follow up questions. So what our practitioners can now do in certain areas of the audit is they can run an AI agent, which will go and review the work they have prepared, and we call that a self review agent. What that does is prepare it, so it gets the first line review from the AI technology to help them identify things they might have missed. So they can then very quickly in real time get that work finalised. So when they do hand it in for marketing to their manager, all that low hanging fruit has been resolved. We should of course, have that manager level review, but by the time it gets to the manager, it's a much higher standard. What that's doing is it's improving the experience of our auditors. Because they're more confident in the work they're handing to their manager. It improves the experience of the clients because they're not getting all those late follow up questions. It's been getting dealt with in real time, and of course it's helping with audit quality.
PL: And less time wasting for the managers? As you say, on the low hanging fruit, errors and missions. So how fast does the technology turn round that first sift?
RH: Minutes.
PL: So completely different to the way it was before?
RH: It's fantastic.
PL: Tell me about how you're introducing it into the teams, in terms of the people themselves.
RH: As you can imagine, this is a big global effort. So what we do before we roll it out, there's a lot of training involved. So because we're rolling out this at an enterprise level across the whole organisation, we have 130,000 people who we're training globally all at once. So we're training people how to use it. Some of the early things we did was train people on responsible AI usage frameworks. So talking about the importance of humans in the loop, taking responsibility for the output that the technology generates, and that's all been really important, getting people used to the technology before we start deploying it at scale.
PL: You must have done a huge amount of hearts and minds work with all your people globally. Enormous project in itself. How did people respond when you first told them this was going to happen?
RH: There's been a lot of jaw dropping when they see the technology working for the first term. What that generally then leads to is quite a lot of excitement in terms of what this means. Working for an organisation that is right at the frontier and cutting edge of this and people's minds very quickly start going to, "okay, great. So does that mean I can move on to the more challenging, insightful, sharp edge side of the profession and does that mean my career is going to accelerate faster? And what does that mean for me in terms of my progress as a professional?"
PL: Must be some nervousness as well though?
RH: Certainly amongst the younger crowd. It's so in the mainstream and in the media and you've got these big celebrities, CEOs running these organisations. I think everyone just really wants to be involved in this.
PL: Up the organisation where they haven't used AI? Because obviously your graduate intake have been using it through university. So they're familiar with AI as a concept and as a tool. But further up I'm guessing not?
RH: Well you always get a level of scepticism with new technology, right? We've had a lot of this over the past years. We have people talking about blockchains, we had all these other things, we had NFTs. So people are used to a lot of technology hype that maybe goes down a bit of a curve and doesn't quite deliver.
PL: So quite a lot of conversations about that?
RH: So you get conversations with "how quickly is this really going to happen? How quickly is this going to be removed?' But then when they see it appearing in the EY canvas files and going live very quickly the imagination comes and they're like, "wow, this is real and this is moving quickly and this is actually really going to drive tangible change in the way we deliver audits."
PL: People must have had reservations?
RH: You get the occasional concern, "what does this mean for me? What does it mean for my job going forward?" Which is the obvious question. I think the answer is you go back to that increasing complexity in the profession and what we do and the opportunity that this creates for us to free up people's time to really do the part of the job that's enjoyable. So that's the challenging management on their key judgements. It's really understanding the business, it's providing those sharp insights and frankly it's accelerating people's careers and getting to that point of the role that's really interesting and exciting. Not that administrative piece.
PL: So you bring them with you by saying, "okay, there's going to be less transactional stuff, less repetitive, dull stuff, more opportunity and there will be training for you around the more engaging and challenging stuff as we go"?
RH: Exactly right. We're training people to use this technology. So I talked about the agentic AI we're deploying within our audit platform, but of course we're also giving people desktop based productivity, AI technologies. So, Microsoft Copilot is a well known version of that and we're training people to use that today to increase their own productivity. So we're giving them this unique combination of skills, AI, how to use AI themselves, how to build their own agents, but also showing them how to do it at an enterprise level, at scale, safely and effectively. I'm not sure there are many places where you can get that experience at the moment. So in terms of getting training to be a professional for the future, it's a really great place to be.
PL: Tell me about the clients. How did you manage that conversation?
RH: So clients are, they're appreciative of us being able to come with a point of view as to how to really do this and to do it well at scale. So you talk about EY's client zero, I don't know if you've heard that phrase, but when we go talk to our clients, we talk about ourselves as client Zero. So we are walking the walk here, we're not just talking the talk. And companies are really interested in how you do that? How do you get to that level where you're ready to start deploying these things?
PL: How long did that take?
RH: So it's been a journey. It's been a journey and there's been a lot of planning as you can imagine. So we've taken a three-pronged approach here. So you've got your technology platform, everything sits in, which is Canvas. So we've had to make sure that's ready and it's connected to all the various different technology assets we use to deliver an audit. You've then got our methodology, which we've been focusing on, streamlining that, making things more standardised, we introduced this concept of guided workflows and that creates an environment and a methodology that AI is better equipped to navigate. Then there's all that people training, I talked about that upfront, making sure people are comfortable with this technology. They know where the responsibility still sits with them, the user, the professional.
PL: So, from start to now is how long? It can't be that long, because people haven't been even thinking about doing this for that long?
RH: Yeah, I mean, a good couple of years.
PL: So not an enormous amount of time?
RH: Not an enormous amount of time. This technology is moving fast, right? And that's what's so interesting, but also so exciting about it.
PL: So how long did it take to get to the point where you looked at these AI agents and how they were functioning and you could confidently tell clients you were going to use them in real audit tasks and try and bring the clients with you on that?
RH: So, like I said, it's been very iterative. So you start with that project management stuff, the research stuff, and now we're introducing them one at a time to do very specific tasks. By talking about that journey and how we've got there, we can build a lot of confidence in what we're doing. You look at the impact it's having on our files and our quality scores, everything's tracking in a very positive direction. I think the really exciting thing for me when we talk to the companies we audit about this is how we can start really using it to sharpen the insight and perspective we drive from the audit process. So I think clients are starting to feel like less of the conversation with the audit team is about sort of administrative staff and following up on samples. And it's more about focusing on things we are challenging on perspectives. We've got things we've observed, sometimes more challenging, potentially uncomfortable conversations, but more intellectually interesting conversations.
PL: So what sort of questions did clients have for you? They must have had a lot.
RH: So we get the questions about the building blocks together there. Because of course you do. You get the questions around security and safety and you know how we governing this, these sort of read horror stories in the press and they Yes, they have questions around, how are you mitigating against that? You get the obvious questions around how are you sourcing the technology, where's it coming from, how are you building it? So there's a whole raft of questions as people explore all of this. We get a lot of technical questions.
PL: Yeah, because then, as you say, risk, it's all over the media. You can't really talk about this without talking about the perception of risk, indeed, the actuality of it. It was last week, wasn't it? We saw the monitoring organisation metre and their report about AI performance across the piece. Honestly they did make you sit up even in a media world where there's a lot to talk about risk. Because there were 44 examples of AI overstepping the mark just between February and March this year. The thing that leaped to my eye was agents routinely attempted to cheat on our hardest evaluation tasks, often in flagrant and elaborate ways that we believe humans wouldn't consider. So there's a kind of, I don't want to use the word deviousness, but a level of imagination that, I think, surprised them how their agents were responding. Tell me, obviously you've thought about this a lot?
RH: Yeah.
PL: It's all going to be about anticipating and mitigating those risks. How do you do that? Is it all about human oversight?
RH: Human oversight is a massive part of it, but maybe if I build towards that. So I think there could be a common misconception here that these AI systems consistently operate with unlimited freedom, and that is not how we have designed these. So our agents are focused on very specific tasks and they operate within very clearly defined boundaries. So they only have access to specific information and tools that are relevant to the task they've been designated. So have you ever heard the term atomic AI?
PL: Explain it for us.
RH: So this is, rather than having an AI agent that is all powerful and can do things, so in theory you could create an AI agent with a chat interface, stick it in our file and give access to everything, you know, "please complete the audit for me." It would offer, operate with a huge level of unpredictability because it would navigate how to do that and it'd be very difficult to test, govern and deploy.
PL: Traceability would be really hard?
RH: Incredibly difficult. So we design very specific agents for specific tasks. They go through a very careful governance and testing process. We'd call it certification. We've got a responsible framework around AI that sits around that. There are nine core principles, if anyone's interested, you could Google them— ' EY's is nine principles of responsible AI.' What that allows us to do is make sure that they are very focused on specific tasks and then you build sort of an orchestrator that sits over the top to get the right agents to do the right thing at the right time.
PL: So transparent, step-by-step process for everything?
RH: Yep.
PL: And the data, the training data?
RH: It's interesting you use the word training, it's a very human word, isn't it? Sort of trained humans.
PL: I think it's how one thinks of it, isn't it? '"What does this AI agent know? We've trained it on such and such data." So yeah, tell us how you would think of it?
RH: We are not training large language models. We're using preexisting large language models and we're putting instruction frameworks and guardrails around them, and we are giving them very specific remits to do very specific tasks. So this is a lot more about how you create that operating framework around the agent than actually training it itself.
PL: Okay, and tell me about people, because I don't want to get into the conversation about, are you losing people to do this? We've talked about that in the podcast before. Obviously we all understand there's going to be stuff around that in due course for the whole sector. But how do people fit into this process in terms of the checking? Are you literally run everything side by side? I'm imagining you have to, don't you, in the early days?
RH: So this is where AI user experience design becomes really interesting and important. And fortunately I've got much smarter people than myself who can help us navigate that. So if you think about an auditor using AI in an audit, the auditor, as we say, needs to be ultimately responsible for its output. So it needs to understand the auditor. It needs to understand what the AI is doing each step of the way. So that's about creating transparent user interfaces.
PL: Yeah, so they understand what they're seeing.
RH: You can understand what it's doing, when it's doing, step by step. A good example might be if you were creating an AI to do a financial statement tie out, so making sure the accounts cast everything ties out. You could write quite a clever prompt that would come back and say what doesn't work? But it wouldn't necessarily visually show you everything it had checked. That does work. So how would you know if it's missed something? So you then get into this user design point around transparency to allow the user, the auditor to ultimately validate what the AI's done effectively. So what we're trying to do is get people to move smarter and faster using the AI, not delegate responsibility to it.
PL: But it sounds like a very people heavy process still, because it has to be?
RH: Yeah. People are absolutely at the centre of everything we do in terms of our audits.
PL: Going forward, how do you see that evolving?
RH: I can't see our people not ultimately being responsible for this in the foreseeable future.
PL: Well, the regulatory environment demands it now, doesn't it?
RH: Yeah, absolutely. You can't delegate authority to an artificial intelligence. I don't think you'd want to. This is again about: how do we make it so the administrative side of what we do is accelerated, so we can then focus on where we can add real value and perspective and sharpen our insights? There are examples already where AI can help us very quickly, research market trends or changes in legislation or unusual anomalies in the data that's in front of us that we're auditing.
PL: Really good for that?
RH: Really good for that. So if we can get our auditors out of the weeds, the AI, doing sort of the transactional routine work still in a transparent way where the auditor can go in and review and validate what it's done, but getting everyone focused on the sort of the more challenging, interesting areas. I think that's going to really elevate audit quality. It's going to improve the experience of our people and it's going to give a lot more value out of the audit process. Then what we haven't touched on is, of course, as companies start moving towards using this, we are going to need to come in and think about, "well, how do we get comfortable if companies AI is doing what they think it's doing? How do we audit the AI's output? So there's going to be a new focus and need to focus on sort of AI assurance and to free up the capacity to do that?" We've gotta get that first bit, right.
PL: So how long has it been since you actually got this up and running in the real world?
RH: It all moves so fast. You've gotta look back and think it's happening quickly as you say. So we rolled out the first agents in our audit files last year in the autumn, and that was the research agents. So that was the, "go help me understand technical accounting, go understand audit methodology". Then in the spring of this year we started moving towards those project management administrative agents. Then throughout the summer, that's when we started releasing the agents that can do the drafting of, or helping us draft audit work. And those review agents I talked about,
PL: I mean it's an unprecedentedly rapid huge change isn't it?
RH: It is, but it's very controlled, as I said, because it's one agent at a time or groups of agents, it's very specific and deliberate. We're not releasing agents that can cover the entire audit file. It's at the moment it's very specific tasks, very specific areas. Then the plan is as we get people used to that, we keep those feedback loops. We learn from user experience, user design, and how to make it effective. The goal is by 2028 to have most areas of the audit, having AI agents supporting us in at least some way.
PL: So as you say, it's iterative, you're presumably tweaking and learning all the time. Big lessons learned so far? Anything unexpected?
RH: The big lesson so far is the value of having conversations outside of your immediate ecosystem or bubble on this stuff. We're doing really well with this, but we're doing that by having lots of conversations with other parties about it. So be that our regulators and taking them for the journey with us, be it the companies we work with or just even other areas of business. I sit in assurance, really interesting hearing what our consulting colleagues are using it for because we're all adapting and learning to use this technology at once. You might have seen in the press that Microsoft and Harvard have identified us, one of the 14 leading frontier firms in terms of this, and that gives us a fantastic cohort to work with and ideate and understand what else is going on in the world. So I think that sharing ideas and talking to others is a huge learning.
PL: Puts you right in the spotlight there, doesn't it as well?
RH: It does, but I think it's a good spotlight to be in. It's really exciting. Everyone wants to be on the frontier of this. Frankly, if I was a young professional coming into the industry, I think this is exactly where I'd want to be.
PL: Now onto Prediction Markets. Podcast regular Polly Tsang is back with us. Hi Polly.
Polly Tsang: Hi Philippa.
PL: Now I'm betting most people have never heard of prediction markets. We're going to get into why they might matter, but first, what exactly are they?
PT: Should we take a step back? So insider trading, why talk about insider trading? So it probably isn't something you associate with the next James Bond, when a CEO retires or when iPhone might release a foldable iPhone. Because insider trading in the past has always been about shares about insider financial information.
PL: People, the city.
PT: Yeah, people in the city who can drive the movement of shares going up or down. But prediction markets are going to change all that. Suddenly, all this information that may just be confidential right now, but can now be monetised, right? Because what prediction markets are essentially is imagine FTSE, but instead of trading shares, you're trading probability of events happening.
PL: Celebrity gossip...
PT: Potentially celebrity gossip. So let's stick with— so, Bank of England—will it cut interest rates next month? Taylor Swift is the first child arriving imminently, but suddenly it all becomes monetisable information. Now, first you might think, oh, this is just a bit of harmless fun, You're betting on "is Idris Elbra going to be the next James Bond", right? But actually, as a company, the reason why as a board member, anyone in charge of corporate governance needs to take heed is because suddenly, confidential information becomes monetisable and insider trading remit potentially can become much, much bigger.
PL: So this is messy, isn't it? Because when is insider trading, insider trading? Because within the definition of legislation and regulatory frameworks, I'm guessing there's quite a specific definition of how that is and what that is. Financial data that only you have access to. I don't know how it works, Polly, tell us, but if I'm betting on Taylor Swift having a baby or not, that's not insider trading? So why do we have a problem?
PT: Well, not right now, but I mean, it is a very interesting question, which actually like where does it fall within? That's very unclear, as in if you're trading information on these markets, under what laws does it fall within? So in the UK they're still clarifying the regulatory perimeter currently, if it's financial, potentially under the FCA, if it's not under the gambling commission. But the FCA has already said that this is an area the perimeter needs to be examined and clarified. But stepping aside, like even without the regulations, this is something that companies need to address now because the risk exists.
PL: There's two platforms, right?
PT: Yes. So Kalshi and poly markets are the two main platforms, and actually they've both exploded last year. We're talking quadrupled, okay? So in part because of the World Cup, because a lot of people have been placing bets on it. Some analysts estimate that annual volumes will reach a trillion dollars by the end of 2030. That's like we're no longer talking about niche websites, we're talking about lots of people placing bets on uncertainty, right? So it's not something we can ignore. Now, an interesting tidbit is actually, you might imagine thousands of people making small bets. But Wall Street Journal analysts have found that actually 67% of the profits have just gone to 0.1% of accounts, which kind of raises the question, right? Is it, are these really good traders or are people trading on information that other people might not have? And I think the recent case of Google has really pinpointed the issues that we're facing. So for those that don't know, the inmate, the US authorities charged a Google employee with using material non-public information to trade on poly market contracts relating to Google's year in search ratings. It's quite funny when you hear about it because this employee under the handle of 'Alpha Raccoon', earned around $1.2 million, allegedly by correctly predicting 22 out of the 23 Google's most popular searches, before those rankings were publicly released. Now, it sounds funny, but importantly, I think people need to remember, this wasn't about earnings, right? It wasn't about a takeover. It wasn't even about Google's share price. It was just confidential internal information that happened to have value because prediction markets existed, right?
PL: How did that come to light? Do we know? Because I'm wondering, I mean how would anyone know if anyone was placing bets on financial confidential information?
PT: That's a really good question because some people, especially the founders of some of these trading arenas, they argue that and I see their point. They argue that actually it's a lot easier, it's a lot more auditable when you commit, say, insider trading on these platforms. Because you can see exactly who made the trade when they made the trade and what events happened afterwards. So arguably, for example, poly-markets use crypto to place trades. You still need to trace it because it's anonymous, right? But really there's a very clear auditable trail. So arguably some types of misconduct may come easier to spot. But I think the wider question here is you are no longer talking about financial information, right? I can give you three really good examples to demonstrate how this might affect a company. It doesn't even have to be in financial services. So the first one, let's start with a bit lighter. So next James Bond, everybody's wondering, right? I mean I'm sure you have an idea. I'm backing a few people, but imagine you're in the room where the casting of the new James Bond. You know who the new James Bond is. The lawyers acting for the actor might know who the next James Bond is. Potentially some of the marketing people might know and they're not going to release this information until next month. But suddenly, by you knowing it, it's no longer a bit of gossip. That's monetisable on the prediction markets, right? You can now place a bet and say, “I know who the next James Bond is and potentially make a lot of money.” Is there an issue with that? Question mark. But if we move that a little bit further, say you told your cousin that, again, Idris Elba is going to be the next James Bond and your cousin places a bet in the poly-market on that. Like, is that still insider trading or is that just like, where do you draw the line? That's the first example. Now let's bring that example from Hollywood back to the boardroom, right? Prediction markets say, okay, company X as CEO will resign by the end of the year. Okay, so you, maybe, the chair knows, the board knows. Maybe HR knows because maybe there's succession planning in place, but the market doesn't know, and technically like that's just confidential information. It's not to do with the earnings. However, that is now monetisable. And I think the final example I'll use to bring this home is say you're deploying an AI product, right? But your engineering team has told you, okay, there's going to be a six week delay. So your engineers know your supply chain might know, your market team might know, but again, if you know it, you can suddenly put that into the poly-market and go, "Hey, actually I'm going to bet like this fantastic new open AI's new model. It's going to be delayed by six weeks." But the conscious of that also is there's an element of corporate risk in terms of corporate espionage risk. Because if someone's betting that, and if somebody's looking at the market and they can see, "oh, somebody's placed a very well timed bet saying that there's going to be a delay, suddenly things that you think are confidential are not, and your competitors will know that. So there's that element as well to consider."
PL: It's interesting this isn't it, because this is accessible to everyone. Back in the day, insider trading, you had to be working in that sector. It was about finance in the city. You had to know a guy or a gal, and that's where you got your information from. This is totally different. This is any organisation with any form of information that isn't known that might perhaps move a market or help your rival. So now we've laid out this potential nightmare, how should businesses respond to this?
PT: Well, I think before we get to that, I think there's also an added twist where it's not just insider information. Now people have the ability to manipulate the market as well, using this. So I'll give you an example. Your CFO is about to do an earnings call tomorrow. You're writing the speech you want to put ‘pineapple’ in that speech, okay? You want to make the CFO say ‘pineapple’. Then you place a bet in one of these markets saying, CFOX company is going to say ‘pineapple’. You make lots of money. As a speech writer, you control that so you can actually manipulate the market.
PL: So you create the outcome?
PT: You create the outcome. So it's not just, "I have insider information," but you have the ability to modify that. Now, back to your question, as a company, what do you do? You do not wait for the regulations to come because the risk exists now, right? So you're doing corporate governance, you need to act now. So I'm going to say five points. First one is awareness. Just making sure that the board and executives in your company are aware that these markets exist because many organisations simply haven't considered it. Second, update your policies. Don't just review insider dealing and conduct policies. You can't just refer to securities trading anymore, right? Consider if any of these policies need to cover prediction markets and any of these event contracts. The third one is to update your codes of conduct. Because this isn't just a legal issue, it's almost an ethical issue in a way. So making employees understand that confidential information may have value, even if it doesn't relate directly to the company's share price and bringing the ethical perspective to that. The fourth is training, and again, that goes back to awareness, but particularly for people who might be involved in areas where this information can be monetisable. So your executives, HR communications, product teams and people involved in major announcements. The last one is just monitoring developments, because this is an area that is developing very quickly. Actually financial institutions have already started implementing policies around it. For example, Goldman Sachs and Morgan Stanley's have both updated their policies. So people are looking at this, but I think people outside of financial institutions need to look at this a lot more.
PL: Do we know how Goldman has dealt with it in policy terms? I mean, have they banned people from using the platforms outright or is it a less defined prohibition?
PT: Yeah, you're absolutely alluding to the right thing. Because it's so wide, how do you do it? So my understanding is some financial institutions have outright banned their employees from using prediction markets. However, there's a question. Well, personal liberties. Also I think you made a really good point that this parallels very well to what companies were trying to do when social media first came into play.
PL: Oh yes . Do you ban all your employees? I mean, some tried, didn't they? And that really went badly.
PT: Yeah. So, it's a really new area.
PL: I mean they felt social media fell back on the don't do anything stupid, be ethical policies, didn't they? I think the policy shrank from five pages to a short paragraph in a relatively short space of time, didn't they? To what worked. That sense of relying on your employees to understand the risk, and do the smart thing, and this sounds like it might have to be the same.
PT: I completely agree, Philippa. I think that's probably right now, anyways, probably a sensible approach before any set regulations, but also because of the massive remit you're looking at as potential monetisable confidential information.
PL: You mentioned regulation regulators are looking at this, I'm guessing?
PT: I don't actually know how closely they're looking at this at the moment. I did see it on their radar as an area of clarification because it's very difficult, right? FDA's remit financial information, but we're talking much, much, much wider than that. If you're betting on things like, "is the CEO going to wear a red tie at the next event?" That's not financial information, so who's going to cover it?
PL: It's not really gambling commission either, is it?
PT: Well...
PL: Or is it?
PT: We don't know. We don't know. It could be, but even if gambling commission's REIT is this, I don't think currently they would cover the aspect where I mentioned where you're making the market yourself. Yeah. That whole point about pineapple when you're manipulating the market, I'm not sure what covers that aspect in the UK with non-financial information, so remains to be seen, but there's definitely something that corporate governance teams can be doing right now, even in lieu of the regulations.
PL: Yeah, and from what you've been saying should be doing, at least having the conversation about it.
PT: Absolutely should be doing, because the question for boards now is not just, “how can we deal with financial information insider trading?" It's like, what confidential information do we have that could potentially be monetisable? It's a much bigger question.
PL: Really interesting. Thanks Polly.
PT: Thank you.
PL: That is it for this episode. Check out the links around the subjects we've covered today. When you go to the show notes to log your listen as CPD, don't forget to do that next time on the podcast. What if Russia invaded Poland? We're taking a hypothetical look at what an expansion of the Ukraine war could mean for British businesses and how they can plan now to combat those risks. If you have feedback or ideas for the podcast team, please do share them with us. Here's the email address podcasts@icaew.com. Meanwhile, if you can find a moment to rate the series on your app, we will be very grateful. Every rating helps us reach new listeners. Thanks for being with us.