Insurance fraud isn’t new and it’s not going away. The financial climate is acutely challenging and unfortunately provides fertile ground for those looking to commit fraud.
Indeed, the earliest recorded instance of insurance fraud goes back to 300BC in Greece, and today it remains a problematical risk, amplified by the on-demand world we live in.
After the internet entered general use in 1993, its impact on society has been unprecedented. It has played a significant part in changing insurance distribution, and increasingly insurers are extending their digital strategies to include online claims fulfilment, featuring straight-through touchless fast-track processing. Among the digital insurance distribution and claims processing are the fraudsters: both the opportunist chancers and the organised.
The maturity of insurers’ strategies to tackle fraud varies across lines of business and global regions, and covers application and claims fraud, as well as fraud in the supply chain. This was reinforced at the October 2023 Global Insurance Fraud Summit in Edinburgh, with representatives from Australia to the UK, as well as at September’s annual conference of the International Association of Special Investigation Units in Dallas, Texas, attended by an insurance industry audience of over 800.
Understanding the level of fraud impacting insurers at an industry level continues to rely on estimates, which are frequently challenged, with understanding at individual insurer level also varying. Neither of these are particularly helpful because, as with any risk, there is a necessity to understand the frequency and impact of fraud to inform the right-fit strategy.
Another key metric often missing from insurer fraud figures are the fraud losses – the ones that got away. The reporting is typically focused on the positive – claims fraud savings – but excludes the fraud losses. Both should be routinely reported on.
From gut feelings to automation
The availability of technology to protect against insurance fraud losses has also changed. The world has developed from ‘gut feeling’ manual application of fraud rules, through to automation of such, then analytics – more recently referred to as artificial intelligence (AI), and now Generative AI. Regardless, whatever technology is employed, when it comes to managing the identified high-fraud-risk claims, computers cannot be used to automatically deny a claim on the grounds that fraud is suspected or identified. That is the role of experienced people, be it claims handlers or fraud specialists.
The earliest recorded instance of insurance fraud goes back to 300BC in Greece, and today it remains a problematical risk
Using the right technology to complement the role of people by automatically identifying those claims where the risk of fraud is highest, and then enabling a consistent process to be followed, drives all important operational efficiency.
At PwC we recognise the tremendous value the right technology can provide, but also the significant shortfalls of using the wrong technology, and that which relies on generic off-the-shelf analytics. The outcome of such is high numbers of false positives, unnecessary operational impact, and erosion of strategies to minimise fraud losses. It also does nothing for managing the technology sceptics.
AI is now part of our lives and often more so than many realise. Combining people and smart technology can be incredibly powerful. However, it’s essential to understand that the value of AI isn’t confined to the technology, rather how businesses use the systems to support and assist people. Detecting insurance fraud is, unsurprisingly, consistently referred to as a use case that is enhanced by adopting AI. But, as with all uses of AI, it is important that guardrails are in place.
PwC has decades of experience in tackling insurance fraud, including for travel and health insurance. This enables us to mitigate fraud risk and help to identify gaps that technology can help with. Three technology enabled tools to support managing the gaps are currently in our focus.
When we consider the opportunities, our first and essential consideration is whether the technology will contribute to accelerating the rapid settlement of genuine claims. There are other key considerations:
• Will it enhance fraud detection?
• Can it be integrated in a digital claim journey?
• Is it ethical?
• Does it fit with legacy technology?
• Is it cost prohibitive?
• What is the resource impact?
Regardless, if the technology conflicts with the first core customer service obligation, then it fails at base one and is excluded.
Claims fraud technology solutions
We have invested significantly in the development of our own new flexible, proprietary, AI-driven, end-to-end, multi-capability claims fraud technology solution, and are now making this available to the insurance industry. Our development approach responded to feedback from clients about their experiences with fraud technology. We chose to develop the solution in partnership with insurers and are continually seeking to identify unique solutions to complement it. We have identified two new solutions that we are now partnered with across global regions, which are simple to trial or deploy, and respond well to each of the key considerations.
The first automatically identifies high risk instances using voice analytics, and AI features in the process. It is not a lie detector, does not use voice printing, emotion, natural language processing, voice stress, tone, inflection or pauses. Nor does it require any personal data, details of the claim or any concerns, or a lengthy telephone call. It is also non-intrusive and can be used in multiple different languages, which is very important for travel and health insurance.
It uses a small number of pre-recorded automated highly customisable voice-based (not a chatbot) questions that only require Yes or No verbal responses from a claimant, with the voice questionnaire completed in just a couple of minutes. The captured responses are automatically used to rate the presence and level of risk at an individual question level. From then it is a matter for people to determine what action to take on the risk identified or not. Not only is the solution unique in what it does and how, but the use cases across insurance extend beyond claims.
The second tool takes it further than identifying risk, to being an evidential solution. It automatically and in near real-time, performs thorough forensic authentication of digital documents and images, presenting the outcome back to the end user. It can be integrated to authenticate every document and image received at scale, and in manual form. The solution has been developed to perform forensic authentication, screening and fraud detection on digital documents and images, using over 100 detection points. It examines documents and images and determines if a file is unaltered or original, an original generated by a specific device, or the result of manipulation through editing software. It can also determine whether the files’ geolocation data corresponds with claim details, and with many other capability features. With its application able to evaluate a vast array of file types received in insurance claims (and beyond), it too brings flexibility.
We have invested significantly in the development of our own new flexible, proprietary, AI-driven, end-to-end, multi-capability claims fraud technology solution
Whenever technology and mitigating insurance fraud is mentioned, the cost to acquire and implement the solutions, and value derived, are rightly the first questions and challenges raised. The insurance industry does unfortunately suffer with the impact of legacy technology, which can have bearing on the adoption of new technology, undermine its value and bring unwanted cost. This is precisely why we work with technology that responds to these challenges, fits with non-legacy environments, and is not cost prohibitive. New and advanced smart technology does not have to come at very high cost, which is important for all lines of business including travel and health insurance, where claims reserves can be at the lower end.
It is important to underline that technology is developing at pace, and arguably never has careful selection of solutions been so important. The wrong selection can result in technology being acquired that rapidly becomes a legacy burden. That will undermine the very best intentions to fight insurance fraud, and the broader business model.
The decreasing pool of experienced and trained people to identify and deal with fraudulent claims was raised at the Global Insurance Fraud Summit. Although on one hand this provides a driver to capture and optimise the experience, expertise, and knowledge to inform AI, it also highlights the challenges in fighting insurance fraud. Smart insurance fraud detection technology is largely redundant without skilled people dealing with the output and making human decisions.