Photos form the backbone of billions of dollars in claims. Yet photos can be so easily and convincingly altered that insurers are having difficulty uncovering the deceit hidden in the altered metadata embedded in images used for claims and underwriting.
Forgeries are subjecting insurers to a growing risk of losses from well-doctored images, especially for auto claims. The potential to use readily available technology to slip altered images past insurers has grown exponentially in recent years. Fraudsters thus are increasingly targeting insurers with doctored images to support their crimes.
The emergence of consulting firms that focus on uncovering photo fakery with insurance fraud and other crimes is the marketplace’s way of confirming the problem is real and growing.
Photos are becoming easier to alter. Even consumers with only basic technology skills can use any of the many photo-editing programs that can create professional-looking altered images to scam insurers during policy transactions.
In an era of low- and no-touch policy transactions and claims speeding through the insurance pipeline, the fraud risk from altered photos is growing more acute. The exposure, so to speak, is especially true with more insurers accepting more photos as part of seamless low-touch claims, where scrutiny can be lower.
Many insurers ill-prepared
Photos submitted by claimants can provide pivotal evidence allowing insurers to verify underwriting and claims for millions of Americans. Yet many insurers today are ill-prepared to penetrate well-doctored images that can be easily submitted.
Using free or inexpensive photo-editing phone apps, fraudsters can easily create or inflate car damage by altering the date or time of a claim incidents such as a vehicle crash.
“The days are disappearing when investigators can know a digital image accurately depicts a claimed scene, person or event.”
Instagram, FaceTune, iPhone apps and other widely available photo-altering programs make it easy for most people to alter photos. Thus it’s a small step for people to skillfully manipulate images for insurance cons.
The days are disappearing when investigators can know a digital image accurately depicts a claimed scene, person or event. Defeating burgeoning photo fakes thus takes on added financial urgency in today’s dawning era of low-touch or no-touch claim transactions up and down the insurance chain.
Even modestly tech-driven consumers can easily defeat insurers with well-doctored photos that support false claims, or doctor bogus policy applications during underwriting.
A revolution in manipulating photos taken by cellphones already is underway at the consumer levels. Basic photo-editing tools are standard software on most smartphones. Hundreds of more-sophisticated apps are easily downloadable for free.
Phone software has “beauty modes,” for example, that can alter someone’s appearance. Different eye color, hair color, facial structure and other elements can easily disguise someone.
Mounting losses from fakes happening?
This tech surge comes at an especially challenging time. Insurance fraud is growing, a majority of insurers say in a recent study by the Coalition Against Insurance Fraud. This crime is an $80-billion annual drain of insurers across all lines, the Coalition conservatively estimates.
Without the right tech tools and human skills to detect altered photos, many insurers can expect large and mounting losses as emboldened fraudsters increasingly submit bogus images that stand a high probability of being accepted as valid. Imagine sophisticated criminal rings manipulating photos for large claim dollars, paying well to hire associates with advanced imaging skills.
Photo frauds are relatively easy for average consumers to use with insurance scams. Several photo apps integrate AI algorithms that allow people to add or remove objects in an image, wipe out the background and enhance the appearance of subjects. Here are just two examples of hundreds of easily available techniques: change or mask my location on Android … adjust time and date after capture.
The commercial design app Photoshop is widely used to alter photos for brochures, personal photo portraits, annual reports and the like. Many people have Photoshop skills. It requires training, though is commonly available. A Photoshop user can cleverly change photos to support false claims such as car or home damage, or add injured crash “victims” to a collision scene. I recently talked to an SIU investigator who said he discovered two Photoshoppedphotos in a claim file.
Automobile scams are a persistent, multi-billion-dollar drain on insurers year in and year out. Auto insurance heavily relies on photos, and thus is vulnerable to fakery. Accurately documenting preexisting damage on applications or crash damage during claims, for example, are part and parcel for auto insurance.
“This fraud potential is magnified by the rapid spread of remote insurer photo inspections …”
Altered photos can support almost any false vehicle damage claim. A free or low-cost phone app lets drivers add or enlarge a scratch or dent with little trouble. This fraud potential is magnified by the rapid spread of remote insurer photo inspections replacing on-site manual reviews.
An uninsured driver who rams a utility pole can make it appear the collision happened after she bought a new policy. Simply change the incident’s date and time on her iPhone or Android, then snap a photo. The metadata now reflects the date she selected. She then changes the time and date back to the correct coordinates afterward. When the insurer asks for photographic evidence, the claimant submits the altered image and the false damage claim stands a good chance of being paid.
Imagine a staged-crash ring makes injury claims for seven passengers in a purported two-car collision. Only the drivers were onsite. Yet the ring gives the insurer photos that show five other people at the scene — all doctored into the image, and making expensive injury claims. As with auto damage and repair claims, photo inspections continue to replace in-person site inspections. Larger insurers, especially, are adopting such claims efficiencies.
The core problem is that photography went digital in a less-complicated time. A digital photo is a collection of binary 1s and 0s that are mutually interchangeable: Alter those 1s and 0s, and you can transform the entire photo.
This applies equally to pixilating the image — what it actually displays — and to altering the metadata inside the image file. Nearly all digital cameras embed additional information within the image file. Most smartphones record the time, date and geolocation of the image capture. Data as mundane as the camera model, aperture and exposure times also are recorded.
This metadata should signify accurate, useful information about the context in which the image was captured. Yet current image file formats have no built-in mechanisms to preserve the metadata accurately. Countless tools are readily available online for anyone to manipulate this information and try to defraud insurers. Worse still, tools to manipulate image metadata are built into photo-storage services such as Google Photos.
Insurers and other businesses rely on employees to make underwriting and claim decisions based on what they see. That’s why manual inspections exist in the insurance world.
“Want the image to appear taken last week instead of today? A few clicks start the scam.”
Want to make an image appear it was taken in Europe instead of the U.S.? It’s easy with a little training. Want the image to appear taken last week instead of today? A few clicks start the scam. Imagine inventing a photo alibi for your home arson or other insurance scam. Advancing technology, along with imagination by fraudsters, make such bogus-claim scenarios an increasing threat.
Why not make a false claim for a $10,000 diamond engagement ring that you say you lost at the beach? Or just go to a jewelry store, try on a ring, and photograph it on your finger (with a generic background). Buy coverage, wait the recommended number of days (easily found in an online forum), then claim the “lost” ring. A photo at policy inception with validated metadata (time, date, location and more) can be crucial and convincing evidence of a valid purchase.
GPS data also can easily be manipulated on mobile devices, especially Androids. It’s undetectable unless more-sophisticated checks and algorithms are run at the point of photo capture in a controlled experience.
Example: Take a photo from a jewelry shop and spoof the address as your home. Same with furniture.
Digital photos speed claims
Compounding insurer loss exposures is the rapid emergence of low- and no-touch transaction processing along the insurance chain. The customer-centric business model relies on heightened speed, efficiency and auto-processing of transactions.
Faster, easier transactions mean happier customers. This is especially true for rising generations of younger, tech-savvy insurance buyers who expect this efficiency. Lightning speed and ease of use are key to attracting and keeping younger consumers whose incomes and buying power are increasing as they move upward through life.
“… altered photos also can face less scrutiny to keep transactions moving fast.”
Relying on digital photos provided by customers can greatly speed up underwriting and claim-processing times. Yet altered photos also can face less scrutiny to keep transactions moving fast.
Insurers catch many thieves by analyzing a suspect’s photo metadata. Fraudsters also can easily foil investigators with a minimum of knowledge about image manipulation. For example, a scam artist wants the time and date in metadata to reflect a vehicle crash at 3 p.m. yesterday instead of a week prior. Fraudsters can easily change their phone settings and snap a photo. The data forgery is undetectable to an insurer.
Claims and SIU teams are highly professional. Yet they cannot work fast enough to manually detect altered images. Investigators and claims staff don’t have time to manually inspect the large volumes of photos racing in for policy applications and claims.
Deepfakes, altered videos next?
Technology for altering videos is fast-emerging as well. They are less of a current threat to insurers today; manipulating photos remains the primary threat. Still, they point to how rapidly and thoroughly technology is being deployed to manipulate reality.
Until recently, only professionals at major photo studios or design firms could alter videos to such impact. Now AI can help nearly anyone manipulate even a video with considerable skill.
Deepfakes are another looming tripwire. They take image altering to a higher level. Deepfakes are totally fabricated videos or images of people or scenes, yet they seem clear and convincing.Deepfakes are widely circulating on social media, and appearing in mainstream media. They are causing confusion and allowing false information to circulate.
“Will the day arrive when average consumers can manipulate videos and deepfakes in fraud schemes?”
To grasp the full power of deep fake photo manipulation, view these images of imaginary rooms and this invented person. We may not be seeing deepfakes in claims — yet. However, the technology’s rapid spread greatly increases the chances that fraudsters will literally invent photos to support false clams.
Will the day arrive when average consumers can manipulate videos and deepfakes in fraud schemes? Imagine a home or business arsonist editing himself out of his security video, or editing in someone who’s taller, heavier and from a different ethnic group.
Seamlessly check photos
Better prevention, not just detection, is the core of catching photo fakes. Being proactive at the point of underwriting, plus receiving trusted, verified and forensically checked images can prevent more scams.
Insurers are checking more photos for deception. The right systems can seamlessly review photos while maintaining convenient policy transactions for customers. Key is to define the “right system.”
Insisting on in-person inspections for every policy or claim — including photos — costs time and money. This also creates creates high-friction points for customers. Delays and repeated requests for information can make customers feel their truthfulness is being questioned.
Virtual photo inspections are gaining a foothold in response. Customers can upload images from their phones at policy initiation, or provide documentation to back up a claim. Virtual inspections are a step forward in fraud detection, yet the images still are not necessarily trustworthy.
An auto policyholder, for example, says she was rear-ended. The insurer asks her to use its virtual-inspection tool to capture damage photos for the claim. Yet she wants to inflate the damage. So she simply searches the internet for photos of the same make, model, year and color of her real SUV — with the same damage she described. Then, she uses the insurer’s software to capture those images on her high resolution computer screen.
There is no easy way the insurer can tell the photos are fake. So her $14,000 damage claim is paid after the virtual inspection, or maybe her expensive SUV is declared totaled.
Controlled capture can auto-detect
Such scams can be auto-detected, even in a low-touch claim environment. Next-generation solutions — for images, at least — employ a technology called controlled capture, working with image forensics. These solutions reduce the fraud risks of virtual and manual inspections, while maintaining a satisfying consumer experience. Controlled capture can be embedded seamlessly into automated claim and detection workflow. It inspects photos behind the scenes as they flow into insurers.
Controlled capture is the ability to permanently gather all data recorded the moment someone presses the camera shutter button. Image forensics then auto-analyze the device and other metrics to verify the data’s authenticity.
Controlled capture makes it impossible to manipulate a photo’s time, date, location or contents. It can detect fraud beforethe claim is paid. Controlled capture can apply to photos at underwriting, point of sale, claims and SIU. Insurers receive trusted, verified images that discover fraudsters and make everyone happier — customer, insurer and its employees.
Using controlled capture and comparing the phone settings to a secure server at the point of image capture is required to understand these manipulations.
Lack of trust in photos is a mounting problem. All businesses that rely on photos are at increasing risk of fraud. It starts with insurance but extends to home sales, shortterm rentals, online marketplaces and more. Photos thus should never be trusted unless they go through controlled capture and forensic checks.
Many insurance consumers withdraw their claim or policy application when an insurer requests a virtual photo inspection. Or they delete suspicious images from several photos before they (or their agent/broker) send them to the insurer. This might include a manipulated image that shows mold in a house that’s in good condition, or jewelry with the store in the background. Controlled capture will discover these photo schemes.
We’ve entered an era when photos are irreversibly fair game for potentially billions of dollars of bogus claims. Photo-altering apps and claimants’ need for speed have let the genie out of the bottle. As more consumers learn that they can use photo-altering apps to convincingly change images to inflate claims, insurer loss exposures will continue to grow significantly now.
In an era when speed is pivotal in resolving claims, insurers must use that same urgency to detect and deter bogus photos.
About the author: Dan Gumpright is Vice President and Head of Insurance at Truepic. He has worked on high-tech insurance software for 10 years, is a regular blogger, and is a speaker at insurance conferences worldwide.