Would I lie to you?

At the heart of many fictional spy drama showdowns is some form of lie detection technology. However, the insurance industry is increasingly using this type of tool when it comes to rooting out suspicious claims.

Like every technology, the lie detector has come on leaps and bounds over the years. No longer is there a need for the subject to be strapped to a machine and asked a series of baseline questions to build a foundation before the real quiz starts.

Manjit Rana, executive vice-president for the UK and Europe, Middle East and Africa at voice recognition analysis firm Clearspeed, told Insurance Times that modern day lie detection technology can deliver tangible benefits for insurers and their clients when it comes to claims.

He explained: “In the past, voice analysis technology required specialist operators and a face-to-face interview, which could last 40 minutes.

“However, the technology can now be used with no impact on the client whatsoever. It is used in banking to identify people to allow them to log in to their accounts – it will compare the voice with base sounds.”

Providing more insight on specific insurance use cases around voice recognition software, Rana noted that insurance customers could be asked three closed questions that have either a ‘yes’ or ‘no’ answer – like the lie detectors of old, today’s analytic tools only need this short response to identify an anomaly.

If the system then identifies that one of the answers supplied is incorrect, then it will flag the policyholder to the respective claims team for further investigation.

Rana continued: “When you ask people questions, the brain will measure the response and that will create clear characteristics in the reply.

“The language or accent does not matter. The questions can be asked and we can identify the risks in the response.”

He compared this process to the “metal detector at an airport”, which alerts security “to the fact that a person passing through the system has metal on their person. It does not say whether it is a pen or a knife – it alerts the security team to make further enquires”.

Case studies

In Rana’s experience, insurance firms that tap into this technology are seeing “a 70 to 80 fold return on their investment”.

One such business looking to benefit from voice recognition tools is insurer Allianz, which implemented this technology in April 2024 within its home insurance claims division.

Ben Fletcher, director of financial crime at Allianz UK, explained that the insurer does not use voice recognition technology for investigation purposes, but for speedier identification of whether there is something awry with particular claims.

He explained: “At Allianz, we are always keen to look at new technology and the benefits it can bring. We are a customer focused organisation and the customer journey is our first concern.

“We have been using [a voice recognition] system for some months and we have found that it allows us to expedite claims. If it highlights that there is no concern, then we can fast track payment to the customer.”

He added that the system has already supported faster claims payments and recorded a rise in the number of claims identified as having a cause for concern.

“This is about first party and opportunistic fraud, rather than third party,” he said.

“We decided to implement the system in our home claims operation as a priority, but we will roll it out to our motor claims and then discuss within the business whether there are other areas it could support.”

However, Fletcher stressed that the technology is not an investigative tool.

“We are not using it to ask probing questions,” he emphasised. “It is another piece [as part of] our broader systems [that] are used for identification, which include analytics, screening, machine learning and predictive models.

“[The technology] is delivering benefits. Identification is one [benefit], but it is also enabling us to speed [up] payments to our customers.”

Claims support

Ian Carman, director and head of forensic and investigation services at loss adjusting business Sedgwick, believes voice recognition technology is a major benefit for the insurance industry – however, Sedgwick does not use it on all claims.

“We have a very narrow use case,” he explained. “We will use it if there is a risk of some concern around a claim.

“It offers [operational] efficiencies, it reduces the costs of claims – which can support cheaper premiums – but most importantly, it aids the customer journey.

“While the system is a real benefit when it comes to identifying potential fraud, it also had the dual benefits of identifying where there is no risk of fraud, so the claim can be settled as soon as possible.”

Sedgwick ran a pilot in Q1 2024 around the use of a new voice recognition system, which Carman said presents a sea change for the way voice analytics can operate in insurance.

“The new system will accurately and quickly turn voice to text,” he explained. “It will use artificial intelligence (AI) to examine the text and identify responses which are deemed to be of concern.”

The technology can pinpoint a number of red flags, but Carman said the system is not simply the arbitrator of whether a claim is fraudulent.

He said: “The system is there to support and guide our investigators, but it not there to lead them.

“We tested the system on a group of claims which we had assessed – some of which we had already identified as fraudulent.

“The system identified double the number of claims than we had previously identified. It is like having an experienced claims investigator in every call.”

Ethical question mark

However, there is an industry-wide debate around the ethical use of this type of technology.

“Ethics are a concern for us,” agreed Carman. “Insurers have the choice as to whether they inform their clients that in the event of a claim they will undergo voice analysis.

“We take the view that the vast majority of claims are genuine and therefore the controls we put in place are based on that assumption.

“The key for this technology is that the customer does not know it is happening and there is no impact in the customer journey.”