The New Refund Scam
A customer orders food through a delivery app. Thirty minutes later, they submit a refund request with a photo showing a crushed box and spilled food. The platform issues a $45 refund. The food was fine.
This isn't new. What's new is that the photo was generated by AI in under 10 seconds. No damaged food required.
Why It's Exploding Now
Image generators have become remarkably good at creating realistic damage photos. A scammer doesn't need Photoshop skills anymore. They type "crushed delivery box with spilled food on a kitchen counter" and get a convincing image that passes human review.
The numbers are staggering:
How the Scam Works
The process is simple enough that tutorials exist on social media:
Some scammers scale this to hundreds of claims per month across multiple platforms and accounts, generating thousands of dollars in fraudulent refunds.
Why Human Review Fails
Platform trust and safety teams face an impossible task. They review thousands of refund claims daily. Each review gets 15-30 seconds of human attention. AI-generated damage photos have become realistic enough that distinguishing them from real photos in that timeframe is essentially guessing.
The tells that used to work no longer do:
What Platforms Can Do
The solution isn't more human reviewers. It's automated detection at the point of submission.
Effective countermeasures include:
The key is catching fraudulent claims in under 3 seconds — before the refund is auto-approved — without slowing down legitimate customers.
The Business Case
For a platform processing 10 million refund claims annually with a 5% fraud rate, even catching half of the AI-generated fraud saves $8-20 million per year. The cost of automated detection is a fraction of that.
How DeFake Solves This
DeFake provides an API that integrates directly into your refund submission pipeline. Every submitted photo is automatically screened for AI generation in under 3 seconds — before a human reviewer ever sees it.
The result: fewer fraudulent refunds approved, lower investigation costs, and no friction for honest customers.