Is ChatGPT an insurance coverage fraudster’s finest pal?

The usage of synthetic intelligence (AI) to facilitate fraud insurance coverage claims has grown over the previous decade, particularly as AI has turn into extra subtle. However the entry of generative AI may drive insurance coverage fraud to ranges but unseen, a digital intelligence professional instructed Insurance coverage Enterprise.

“With the expertise that’s being constructed now, it’s straightforward for simply the common one who desires to magnify their declare to take action, in addition to organized crime teams that need to make a bigger play throughout a number of firms on the identical time,” mentioned Joe Stephenson (pictured), a former particular investigator and present director of digital intelligence at claims expertise agency INTERTEL.

The issue is that generative AI, comparable to OpenAI’s ChatGPT, has made warping picture proof – a key mechanism to confirm claims – extraordinarily straightforward and fast to do.

“The query turns into which is worse: a thousand people making false $100 claims, or an organized crime group making $50,000 in a single fraud declare?” Stephenson requested. “It simply opens the door to issues.”

Fraud claims simpler than ever?

Photograph manipulation apps have lengthy been used as a software to commit insurance coverage fraud. As these apps grew to become more and more accessible, fraud proliferated.

“The apps the place you possibly can add an image of a automobile and modify injury, comparable to put a crack within the windshield, then ship it out to an insurance coverage firm and get $400 to switch the windshield that was by no means damaged – these have been round for nearly 10 years now,” Stephenson mentioned.

“Superior expertise is making it really easy for the typical particular person to have the ability to have interaction in a sure degree of fraud.”

“Criminals are so fast to leap on this and begin to leverage it of their fraud, that now we’re on this rush to attempt to compete towards it,” Stephenson mentioned. “Generally we don’t even acknowledge the fraud but till we’ve large losses.”

Whole losses as a consequence of insurance coverage fraud within the US have risen to $308 billion in 2022, in accordance with the Coalition Towards Insurance coverage Fraud.

Throughout the border, the Insurance coverage Bureau of Canada estimates that insurance coverage fraud prices in Ontario alone vary between CA$770 million and CA$1.6 billion a yr.

Might generative AI gasoline an increase in artificial identification fraud?

One other worrisome improvement introduced on by generative AI is the rise of artificial identification fraud, in accordance with Stephenson.

Such a fraud, extra generally seen within the banking and finance industries, happens when people mix actual and pretend identification info to create accounts or make purchases.

“Artificial ID is advancing at a a lot sooner charge,” Stephenson mentioned. “Previously, it sometimes concerned an actual one who didn’t have a lot of a footprint, say, any person who has immigrated to the US.”

People would possibly get pulled in by organized crime teams who use their actual social safety numbers with false info, comparable to date of start, to make fraud purchases. Alternatively, fraud actors can cobble collectively completely different items of private identifiable info from actual folks to create “Frankenstein” identities.

However with generative AI, criminals don’t want actual people to create artificial IDs. Instruments like ChatGPT can generate reasonable photos of individuals utilizing current knowledge, and with a larger proportion of insurance coverage transactions taking place purely on-line, verifying actual identities has turn into an enormous problem for insurance coverage firms, in accordance with Stephenson.

“We’ve made it a lot simpler to do every thing with out that bodily interplay that now artificial IDs are build up faster,” the intelligence professional mentioned.

“Generative AI permits me to create a pretend human being. I can pull a picture of an individual that exists, create a voice for them, and construct an entire persona on-line of any person that doesn’t exist, after which use that persona to commit crime.”

Stephenson mentioned there could also be as many as 20 completely different packages out there that people – by no means thoughts organized crime teams – can entry to create artificial personas at scale.

Combating AI-powered fraud with AI

Alternatively, insurance coverage firms can even wield AI within the battle towards claims fraud.

Consultants have identified that by analyzing claims knowledge, AI can establish suspicious patterns and anomalies that might point out fraudulent exercise, serving to carriers scale back losses as a consequence of fraud.

“We’ve received to decelerate,” he mentioned. “When it comes to fraud, it’s essential take into consideration how any person can manipulate this expertise. You would possibly change your course of since you assume you’re going to get a ten% income improve, however are you additionally going to see a 12% income loss since you’ve made it that a lot simpler for a fraudster to return in?

“Authentic policyholders make up most of insurance coverage claims. Fraudsters are a really small half, however a fraudster can have an enormous affect in your backside line in case you make it simpler for them. So, decelerate and perceive the expertise earlier than you flip that change.”

Do you assume generative AI may have a major affect on insurance coverage fraud claims? Share your ideas within the feedback.