A damage photo used to be hard to fake. Today, anyone can add a dent to a car, enlarge a water stain or generate a convincing image of a broken item in minutes. Claims teams now see two kinds of manipulated photo:
- Shallowfakes: real photos changed with an ordinary editing app, such as damage painted on, a date changed, or a photo of a different car.
- AI-generated images: pictures created or heavily altered by an image generator, sometimes from a single real photo.
Checks that help
- Look for reuse. A reverse image search can show that a "new" photo already exists online or in an earlier claim.
- Check the light and the edges. Edited areas often have shadows that point the wrong way, reflections that don't match, or edges that are too smooth or too sharp.
- Watch for generator traces. AI images can show repeated textures, warped text and logos, impossible details, or surfaces that look painted rather than photographed.
- Read the metadata, carefully. Camera details, capture time and location can support or contradict the story. Missing metadata proves nothing on its own, because messaging apps and screenshots strip it.
- Look for Content Credentials. Some cameras and apps now attach Content Credentials (the C2PA standard), which record how a file was made and edited. Most claim photos don't carry them yet, and a valid credential shows the file's history, not that the loss happened.
- Compare with the rest of the claim. Dates, weather, repair invoices and earlier photos should tell the same story as the damage photo.
Don't decide on one signal
Every one of these checks can be wrong on its own. Genuine photos get compressed, cropped and forwarded, and that can look like editing. Treat findings as reasons to look closer:
- Record what was found and where.
- Ask for more evidence, such as additional photos, video or an inspection.
- Route strong findings to your special investigations unit (SIU) with the evidence attached.
- Keep the decision with a person, so valid claims aren't denied on a false alarm.
How Trustable helps
Trustable checks damage photos for edits, AI-generation traces and gaps in camera data, and checks repair invoices and receipts for edited figures. Each finding is marked on the file with its strength, for your adjuster or SIU to review. No AI is 100% accurate, so the final call always stays with your team. See AI-generated claim photo detection and the sample report.