Skip to content
Submit incident
AII-1002-13007
ReportedReported by fewer than three independent publishers, with no official finding.

AI-Generated Fake Product Images Used for E-Commerce Refund Fraud

Since August 2025, consumers in China have used AI image generation tools to fabricate product damage photos and submit fraudulent refund requests to major e-commerce platforms. Testing by a news reporter found that merchants and platform systems frequently failed to detect the AI-generated fake images and approved refunds, causing merchants to lose both goods and shipping fees.

30 August 2025 Β· Pinduoduo Β· Fraud, scams and manipulation
30 August 2025
AII-1002-130072 sources ↓
Quick facts
Category
Fraud, scams and manipulation
Country
China
Source
news.cctv.com
Date
30 August 2025
Drafted by
Raidu Incident Detection

What happened

Since August 2025, consumers have used AI image generation tools to create fake photos of product damage and submitted them to e-commerce platforms to request refunds without returns. The AI-generated images typically added visible damage such as cracks, stains, or mold to photos of intact products like socks, cups, toys, scissors, face masks, and clothing.

A reporter tested AI image generation on four products across Pinduoduo, Taobao, and JD.com, generating damage photos in approximately 20 seconds. After removing watermarks, the reporter sent the images to merchant customer service claiming product quality defects. Merchants failed to identify the fabricated nature of the images and most proactively offered compensation or replacement. Pinduoduo's automated system also independently approved the refund request.

Reported impact

Merchants face losses in both goods and shipping fees when refund-without-return requests are approved. The reporter achieved a 75% success rate in obtaining merchant compensation without detection. Merchants report difficulty and high costs in training customer service staff to identify AI-generated images.

Index analysis

The incident exploits a gap in e-commerce platform refund policies and merchant verification capabilities. The low technical barrier for AI image generation (20 seconds) versus the high cost and expertise required for merchants to detect fabrication creates an asymmetry that facilitates fraud.

E-commerce platforms that approve refunds without adequate verification of image authenticity enable the fraud. However, platforms also face pressure from their own rating mechanisms, where refund dispute rates affect merchant credibility scores, incentivizing merchants to accept fraudulent claims rather than dispute them.

Governance controls

In April 2025, major platforms including Pinduoduo, Taobao, JD.com, Douyin, and Kuaishou adjusted their refund policies, moving from platform-enforced 'refund-only' mechanisms to merchant-controlled handling. However, consumers can still request refunds without returns through merchant negotiation.

Classification

Category
Fraud, scams and manipulation
Industry
Logistics, wholesale and retail
Country
China
Harm status
Actual harm
Intent
Intentional
Timing
Post-deployment
Harm types
Economic or property
Harmed
Business
Failure modes
Deepfake image, Misidentification, Manipulation or dark pattern
AI systems
AI image generation tools

Sources and evidence

This record was researched and written by the Index, based on the following original reporting.

news.cctv.com
用AI做"ζ―ζŸε›Ύ"η”³θ―·"δ»…ι€€ζ¬Ύ"οΌŒζ˜―ε¦ζΆ‰ε«ŒθΏζ³•
31 August 2025
wap.stockstar.com
用20η§’η”Ÿζˆηš„AI图片 ι’»δΊ†ζ‰€ζœ‰η”΅ε•†εΉ³ε°ηš„ι€€ζ¬ΎζΌζ΄ž-θ―εˆΈδΉ‹ζ˜Ÿ
1 September 2025