In Focus: Metadata knowledge key as scammers turn to AI for motor damage fraud

With the fraud space constantly changing, it is crucial that insurers remain aware of the ebbs and flows of such activity – especially in the motor sector.

For example, a look at data published by the ABI earlier this year (24 August 2023) shows that motor insurance made up the largest volume of fraud cases in 2022, with some 42,500 motor scams being uncovered last year.

According to the figures, this represented 59% of total insurance claims fraud.

And with artificial intelligence (AI) having been on the rise in recent times, one such risk for the motor sector is the potential uptick in fraudulent activity arising from more empowered organised crime groups.

Indeed, back in March 2023, cyber security firm Darktrace reported the emergence of more convincing and complex AI scams.

And speaking during an Insurance Times Fraud Charter roundtable event earlier in the year (16 May 2023), Paul Holmes, partner at legal services firm DWF, said computer programmes could soon become the main way to perpetrate fraud during a discussion about AI.

“From a fraud perspective, the worrying thing is we are genuinely very, very close to a period where you won’t need warehouses full of people in whatever country to perpetrate frauds, you can just set a computer programme to do it,” he added.

Digital damage

For the motor sector, one trend that seems to have recently emerged amid the rise of AI is fraudsters moving into scams related to the repair and replacement of damaged vehicles.

For example, Aviva revealed earlier this year (27 November 2023) that it detected 22% more fraud on claims for damaged vehicles last year, of which nearly two-thirds was organised fraud committed by third parties.

Sam Lee, fraud risk and technical manager at Markerstudy, highlighted that in the past, people would stage accidents to commit damage fraud, although said that more recently criminals were moving more towards technology.

“We have seen varying degrees of sophistication when it comes to digital damage,” he said.

For example, Lee highlighted that there were some people who would take a photo of a damaged car from the internet and change the numberplate.

However, he warned that there were criminals using software to add damage into undamaged cars – and that the use of AI was making the images more convincing.

“It has become more advanced in 2023,” he said.

“The older versions of it I have seen – they stood out like a sore thumb. Now detecting it, however, is more challenging because AI is doing a much better job than somebody who has basic skills.

“With advances in technology and AI, it is becoming easier and easier – you can just type in ‘dent’ and the software will put a very convincing dent, generated by AI, into the image.

“That has been quite a big change in the motor landscape.”

Metadata

With AI becoming more prevalent in scams, more insurers have moved to strengthen their fraud preventions measures.

For example, earlier this year (27 November 2023) Axa UK said there was a “growing need for advanced technology to counter evolving fraud”.

And on 8 December 2023, Allianz UK said it had developed a new machine-learning tool, Incognito, in a bid to tackle fraudulent claims with AI.

However, Neil Jones, head of fraud at Verisk, said that when it came to motor fraud, catching scammers was a bit like Whac-A-Mole given advancements in technology.

And when it came to more advanced digital damage fraud cases, Lee said it was on his “risk list” and it was key that counter-fraud experts understood metadata within the imagery.

This is information that helps describe and explain other types of data.

“We have done a lot of awareness work with our engineers as to what to look out for,” Lee said.

“For example, you could look at if the metadata is missing or if it all matches up. If something does not look right, it probably isn’t.

“We are working on scaleable solutions to it – in a perfect world, you would want to be able to run every image that you ever get through something that is going to tell you whether or not it has been aletered.

“That is something we are exploring.”

Jones added that insurtech Verisk had been working on anti-fraud products that could help insurers pick up on metadata issues.

“The image metadata can be checked via an application programming interface (API) to see if it has been altered or to see if there is a discrepancy between the location of an image and the actual location of the accident, or the date of the accident compared to when the image was taken,” he said.

“So, metadata is something that we actively work on.”

Collaboration

While the examination of metadata is key, Jones also felt that it was essential that firms had the most up to date tools, such as a predictive modelling platform to identify new fraud and evolving trends.

As part of this, he felt it was essential that the sector was “sharing intel” so all firms could better work together to combat fraud.

There are examples for collaboration in the fraud space – for example, at a previous Fraud Charter event, Faye Fishlock, head of defendant services at Carpenters Group, cited that two insurers in her network operated a reciprocal information sharing arrangement around spotting potential spoof adverts.

And at the Fraud Charter on 20 September 2023, Carpenters Group director Donna Scully said: ”I have always felt that, over the years, [fraud] is the one area where we put down our competitive tools for the greater good and say, we are going to work together.”

Jones said that from the insurance industry’s point of view, firms will “never sit tight” and always look at what tools are out there to better fight fraud.

“It is a case of insurance companies always trying to be one step ahead of what the fraudster is thinking,” he added.

“If we are working on technology, if we are sharing ideas and thoughts and if we are sharing intel, from an industry point of view, we are best placed at that point to help combat insurance fraud.”