Photo-based claims estimates have been a growing trend the past few years in the insurance industry, and once the pandemic hit, that trend accelerated significantly as the need for remote claims estimates increased. Photo estimates for auto claims doubled early in the pandemic, and Verisk saw a 933 percent increase in usage of our virtual estimating tool, ClaimXperience.
These estimates seem like a win-win for insurers and customers. Policyholders can get faster, simpler service when reporting losses and reduce expenses without the need for onsite estimates. But for fraudsters, photo-based estimates are low-hanging fruit for scamming carriers.
In this series of blog posts, we’ll examine top schemes in digital image fraud and innovative techniques insurers can use to detect each of those schemes.
Re-using prior-loss images
One common scheme we’ve seen fraudsters use for years is submitting a photo from a prior loss as new damage for a claim. This scheme is likely to go undetected, especially if the claimant switches carriers, so there’s no record of the prior-loss images.
But it’s not just claimants that perpetrate this tactic. Some third-party suppliers such as independent adjusters, repair contractors, or auto body shops might re-use images to inflate bills or cut corners on providing estimates. For example, we helped a carrier investigate an appraiser’s claims and discovered he submitted 170 duplicate photos over two years, impacting claims with over $1 million dollars in indemnity paid out.
Using photos from the Internet
Another tactic fraudsters use is downloading a photo of damage from the Internet and submitting it with a claim. With no adjusters visiting a property or vehicle to verify the damage, this is another scheme that can go easily undetected.
For example, this roof damage photo was submitted with a claim in August 2017, and the insurer paid $8,600 for the loss. However, the image was from the Internet, appearing on three different websites, uploaded in 2010 and 2015, years before the date of loss.
Uncovering fraud with binary forensics
Fortunately, there are ways insurers can help detect these schemes. Digital media forensics use AI to identify fraud and manipulation in digital media files such as photos and documents. There are numerous categories of forensics that are applicable to detecting fraud.
Binary forensics are the simplest form of forensics. They are often effective at determining if an image is fraudulent or not and are useful at determining if a loss image is re-used or sourced from the Internet. The technique involves analyzing image pixels to determine duplicates from a database or the Internet.
Challenges in implementing binary forensics
While binary forensics are usually the simplest type of forensics, there are challenges in developing and implementing them. First, it requires expertise in specialized image processing. It’s not enough to analyze image file names to find duplicates. Binary forensics requires algorithms that analyze an image’s pixel values.
Second, it requires a fair amount of computing power for algorithms to quickly search a large database of existing images to find a duplicate. And lastly, you have to be careful of false positives. Effective binary forensics must limit the percentage of false positives so investigators aren’t buried with false leads.
The truth is that most individual carriers aren’t equipped to handle these challenges to implement binary forensics. That’s why it takes an industrywide approach to fight digital image fraud.
Collaboration is critical to fighting fraud
Digital image fraud can be difficult to detect because fraudsters can stealthily move from carrier to carrier. That’s why Verisk is tackling the problem broadly.
We’ve created a digital media contributory database as an enhancement to ClaimSearch®, the industry’s largest loss history database. Participating companies can now view prior-loss photos on match reports, and our Image Duplication Check feature uses binary forensics to determine if any images from your new claims have duplicates in the system or appear on the Internet. Our algorithms analyze each image’s pixels, limit false positives, and help insurers fight digital image fraud at scale.
Photo-based estimates are here to stay. It’s critical to have technology in place to help ensure you’re not paying for fraudulent losses.