This is the first in a series of articles exploring the latest trends in digital media fraud and strategies to catch digital manipulation in an ever-changing field.
The rapidly evolving tactics of insurance claims fraud, increasingly powered by AI-assisted manipulation of digital media, require claims professionals to continually expand their knowledge. Verisk trains its Digital Media Forensics tool on real-world data, helping identify emerging manipulation techniques.

Individual adjusters only see the claims in front of them; check prior-loss images to verify new photos as original.
Here are some of the latest learnings from the field on how fraudsters use manipulated images and documents to support false or inflated claims—and how to spot the deception:
Reused photos
- What it looks like: The same photo submitted on more than one claim.
- Why it works: Individual adjusters only see the claims in front of them, and a small edit makes a photo appear new.
- What gives it away: Matching images across claims (the same dent, stain, or background), identical photos appearing under different dates, loss locations, policies, names, etc.
- Where you’ll see it: Auto: the same vehicle damage photos used for different losses. Property: the same roof or water-damage image attached to a new claim.
- Action Plan: Search prior-loss images before you accept new photos as original.
Photos pulled from the internet
- What it looks like: Damage photos taken from stock libraries, social media, news coverage, or listing sites and presented as the claimant’s own.
- Why it works: Online images look authentic, and reviewers often aren’t searching each one on the web for matches.
- What gives it away: Unusually polished composition for a claimant photo, a scene that doesn’t match the described property or vehicle, and a reverse image search that finds the photo online.
- Where you’ll see it: Property: a dramatic storm-damage image that turns out to come from a news story. Auto: a severely keyed/scratched car that turns out to have been posted online several years earlier as vandalism, then submitted as damage.
- Action Plan: Run a reverse image search on suspicious photos to find potential matches.
Date and location mismatches in metadata
- What it looks like: A photo whose embedded data (capture date, time, geolocation, device) doesn’t match the loss details.
- Why it works: Metadata is invisible on screen, and reviewers don’t always know how to verify.
- What gives it away: A capture date before the reported loss, geolocation pointing somewhere other than the insured address, and editing software listed in the file details. But a cautionary note—missing metadata is only a potential indicator for further review; metadata can be stripped by apps, platforms, messaging tools, and by file compression or transfer.
- Where you’ll see it: Property: roofing and siding photos taken before the reported storm date or at a different address. Auto: a photo of a cracked windshield that was taken before the policy was purchased or changed coverage.
- Action Plan: Compare the capture date, location, and device information against the policy or loss report.
Pixel-level edits
- What it looks like: A photo retouched to add, enlarge, or hide damage.
- Why it works: Modern editing tools produce more believable results, and the changes are small enough to pass at normal viewing size.
- What gives it away: Repeated or cloned patterns, edges that look too smooth or too sharp, mismatched lighting or shadows, and inconsistent texture or noise across an image.
- Where you’ll see it: Auto: scratches or dents that appear to be added or removed. Property: water stains or cracks that appear painted on. Workers’ compensation and medical: images supporting an injury or treatment that look altered.
- Action Plan: Examine suspicious areas for inconsistencies, including displaced objects, unnatural edges, inconsistent textures, or areas that appear copied or spliced.
Altered PDFs
- What it looks like: Invoices, estimates, or medical records modified to misrepresent information, such as a changed invoice amount.
- Why it works: A PDF looks final and trustworthy, and free online editors make edits quick and clean.
- What gives it away: Overlapping or stacked text objects; creation and modification dates that don’t align; a file created by an online editor rather than the issuing business’s system; and a version history that shows earlier values.
- Where you’ll see it: Auto and property: repair invoices or estimates with inflated totals. Workers’ comp and medical/long-term care: altered treatment records or billing documentation.
- Action Plan: Look for suspicious white space, inconsistent values, or calculation errors in totals. Verify information with the original source.