EXCLUSIVE: “Weeding out Fraud” – Tom Clay, Covéa in ‘The Insurtech Magazine’

Covéa’s Tom Clay says that while economic hardship provides fertile ground for insurance scammers, the use of AI can help combat some concerning trends

Economically difficult times inevitably influence the number of individuals who turn to fraud and the ‘victimless’ crime of fleecing an insurance company spikes. In Sweden, the industry estimated that scams accounted for between five to 10 per cent of the SEK 70billion (£5.1billion) paid out to non-life claims in 2022. The Coalition Against Insurance Frauds in the United States, meanwhile, calculated that fraudulent claims have increased there by 285.75 per cent since 1995 and cost an eye-watering $308.6billion annually. In the UK, insurers believe they’ve been losing upwards of £1.2billion per year since 2019, and that identified fraud only accounts for a third of total cases.

But since inflation started rising and the cost-of-living crisis took hold those statistics have shot up – by 66 per cent since March 2022.

This is a jump so significant that the City of London police launched a national campaign in July 2023 to help businesses better acquaint themselves with fraud types in order to prevent them. Those working in the insurance industry describe their side of the challenge as three-fold: they must obviously combat the rising tide of opportunistic fraud, driven by financial hardships. But they also need to find and thwart attacks from increasingly sophisticated scammers and criminal syndicates who are perpetrating digital fraud by taking advantage of the growing number of loopholes, or ‘blind spots’, found in areas like data analysis and third-party connections.

Last but not least, insurers find themselves faced with the mammoth task of policing the digital frameworks that businesses, including their own, have spawned since the digital acceleration post-2020, and which create an abundance of new data and touchpoints – each, sadly, an opportunity for fraudulent activity.

“Claims provide the perfect storm arena,” says chief data scientist for Covéa, Tom Clay. “It has always been where the most creative fraud generally crops up, but it is also one of the hardest areas to get data on to solve the problem – because fraudsters constantly change their tactics, which are really complex things to model. “The structure of the claim itself – long and complicated with a multitude of touchpoints and the involvement of many people and systems – allows for a huge spectrum of fraud types to proliferate by design.”

Covéa Insurance is the UK subsidiary of Covéa Group, a European leader in insurance and reinsurance. It offers a wide range of home, motor and commercial insurance.

“With this technology, we have proven a way that AI can work harmoniously with human expertise to develop a ground-breaking approach to fraud”

Clay’s job is to ensure the firm and its underlying brands gain the upper hand on claims fraud in all of its forms, which he and his team are increasingly doing with AI tools.“When our journey into AI began, the primary focus was on security and pricing,” he says.“As a regulated business, we needed to protect our customers against fraud because, if we could protect them and pay claims where we should and avoid paying those we shouldn’t, we could provide a higher quality service.

”But, as we learned more and more about how to engage people, build products, and experiment on an enterprise scale, we started to move further into customer experience.”

Today, Covéa is trying to apply AI to as many aspects of the customer journey as possible and the company is building its own internal suite of AI tools. Ghostbot, for example, augments processes in order to detect obscure patterns.

“Ghostbot is unique because trust in the AI has been woven into the solution from the beginning,” says Clay. “Every prediction is fully explained in terms that assessors, regulators and customers can understand. With this technology, we have proven a way in which AI can work harmoniously with human expertise to develop a ground-breaking approach to fraud.

“AI has already saved the firm money and in 2022 won Covéa The Insurance Times Claims Excellence Award for Fraud Solution of the Year.

Tellingly, the citation read: “Covéa is genuinely invested in their customers, detailing the steps they take to do their utmost to support them throughout the claims process, with high-level expertise and flexibility in particularly challenging times.”

ASSESSING THE FUTURE

Clay is hoping that, going forward, the firm will be able to further combine external data points and a new wave of internal ones to create more next-generation services that monitor, predict, and actively steer policyholders.“But we want to use AI to augment our people processes rather than replace them,” he stresses. “For example, when a customer calls us after an accident, which is often a time of stress, we want AI to assist and feed us as much data about the problem that the customer is having as they call us. It can help us understand what happened, what we can do, and what the next best course of action is.”

The extent to which AI could transform the relationship with customers is difficult to predict. But it promises to provide intelligent protection that could, ultimately, save not just money, but lives.

“In our business, we’re always hypothesising on how AI could be used, such as anticipating risks before they even occur – like house flooding, or a car crash – which I believe we could do by marrying a variety of freely open data to new customer data,” says Clay. “Take the example of adding connected vehicle data to motor insurance claims, which could give us more than we have ever known about driving.

“Early telematics policies looked at how fast drivers were going and how hard they were braking, but now we can tell at what temperature their air conditioning was on, or whether they had the windows open. And when you start to correlate that with the weather, you could begin to question if maybe the driver had been a bit drowsy.“We are not there just yet, but this is very much the direction I believe we, as an industry, will move in. “It is a natural progression because, with the amount of data we can amass and analyse, you not only get to a point where you can price a service properly, but you also start to know your customer a little bit better and, eventually, can predict how they are going to behave in a number of areas.”

Clay and Covéa champion the power of AI as a money- and time-saving tool, but others are wondering whether the collection of ever-more personal information on people has gone too far. As it stands, studies are showing that most people remain to be convinced of AI’s reliability to establish social reactions correctly or generate conclusions without bias or mistakes, and most are calling for appropriate policing.

Both the EU and UK governments have begun rolling out their respective strategies to guardrail the use of AI by 2025, like approving AI systems fitted into transport and banning some types of AI that present an unacceptable risk of bias because they classify people based on behaviour, socio-economic status or their personal characteristics.

PLAYING CATCH-UP

Clay believes the successful adoption of AI in insurance comes down to how governments and underwriting entities like Covéa work together.

“The barriers aren’t around technology now; it’s public awareness and regulation,” he says. “The public knows that AI has kicked on a lot in the last 20 years and are starting to think about how the technology can help them in their financial life. But there is still a way to go. “And we know that the regulators are now really working hard to catch up, although I don’t think they ever will, truly.

“You could use AI to automate pretty much every single process you want, but is that the right thing to do? And how do you make that transparent and explainable? “So, I think the future of insurance and banking, and so many other areas, will depend on what the regulatory environment looks like and how people understand AI, regardless of the technology that sits underneath it.”

This article was published in The Insurtech Magazine Issue 10, Page 17-18