Generative AI has been used to fight back against
Catching fraud requires a workflow that includes machine learning, predictive modeling and rules engines, states Kimberly Harris-Ferrante, distinguished vice president analyst, Gartner. Gen AI is useful because, she says, “you’re going to package multiple technologies to do the analysis, the workflow, the decision mean, the calculation of risk. And because these technologies are new, they’re more simplistic in how they’re being spun in the industry.”
Insurers are becoming more concerned about Gen AI being applied to create fake images or content that can be used in claims, but their use of technology still has to catch up, according to Harris-Ferrante.
“A lot of insurance companies don’t have modern fraud solutions. They’re not running these state of the art AI-based fraud solutions,” she said. “Mainstream insurance companies are still running some rules that a special investigation unit built, that could be running internally. We haven’t seen any best in class yet.”
Rather than using Gen AI to manipulate damage photos for claims, a bigger issue is its use for false identification of policyholders making claims. Synthetic IDs, which are fabricated identities that are not real people, are being used to make claims, as Karen Jennings, a special investigations unit manager at American Family Insurance, explains.
“They’re misrepresenting their information to the insurance company to take out a policy, and then from there, depending on the carrier, either the claim is paid or they catch the red flags that it’s not an actual person,” she said. “Each case is a different scenario.”
American Family’s fraud investigators themselves use Gen AI daily to detect fraud, according to Jennings.
Gen AI is also being used to get a claimant identity through social engineering, according to Jessica Groopman, senior innovation fellow at Intentional Futures.
Christophe Testi chris@creatives
“There are a wide range of different ways that we’re starting to see this take shape. Some of them are akin to general cybersecurity threats, where AI is used to manipulate people to give up sensitive information or take actions that would commit fraud, instigated by the AI,” she said. “There is even automated social engineering, where AI can be used to scrape public information to generate highly targeted, highly convincing attacks that are targeted specifically to you and your fears, desires, needs in this particular moment, context, workspace, boss or whatever the case may be.”
Small businesses are more susceptible to this risk, both because they have less defensive resources and less insurance coverage, according to Groopman.
“There are ways in which AI is used to combat cybersecurity threats and to create net new threats,” she said. “That is just as true with the use of AI in fraud as a specific cybersecurity use case.”
