Retailers Face Surge in AI-Generated Return Fraud

Retailers are reporting a sharp rise in return fraud that uses generative AI to fabricate evidence, according to Retail Gazette's coverage of data from fraud-prevention platform Forter. Forter's data identifies AI-generated damage claims - fake photos of "damaged" goods and doctored receipts - as the fastest-growing form of return abuse. Retail Gazette reports that 53% of merchants are seeing "wardrobing" (wearing items before returning them), 44% of UK businesses say returns and refund abuse affects them, and almost half of retail leaders surveyed considered scaling back operations or closing this year because of returns pressure.

Some fraud rings have industrialized the tactic by selling "returns-as-a-service," helping shoppers secure fraudulent refunds in exchange for a percentage. PYMNTS documents merchant incidents in which submitted damage photos carried AI watermarks, and, citing Riskified analysis, reports that refunds amount to roughly 1% to 2% of total sales in the dataset analyzed, with nearly one in four refund dollars linked to abuse. PYMNTS frames this against industry figures of nearly $1 trillion in U.S. merchandise returned in 2024 and an estimated $200 billion a year spent recovering value from returned goods.

Editorial analysis: this is a clear, operational example of generative AI lowering the cost of an attack. Synthetic damage images can be photorealistic, stripped of metadata, and produced at scale, which defeats detection heuristics built around low-resolution or obviously edited images and simple receipt OCR. High-return categories such as fashion are most exposed because returns already erode margins, so fraud-driven losses hit profitability directly.

Editorial analysis: defending against synthetic evidence pushes fraud teams away from single-signal rules toward multi-signal verification that combines image forensics, device and identity signals, and logistics telemetry such as delivery confirmation and timestamps. The hard trade-off is false positives: aggressive flagging protects margin but adds friction for legitimate customers, and adversaries can iterate new synthetic outputs faster than static rules adapt.

Editorial analysis: watch for wider adoption of image-forensics and cross-signal verification in returns workflows, more reporting of AI-watermarked or machine-generated indicators in submitted claims, the growth of third-party "returns-as-a-service" marketplaces, and whether industry groups or regulators issue guidance on synthetic-evidence fraud.

Several headline figures come from a Forter survey relayed by Retail Gazette and from Riskified analysis cited by PYMNTS; survey-based and vendor-sourced statistics vary by sample and method, and the topline returns figures are industry estimates rather than audited totals.