AI-generated images of “damaged” goods are fueling a sharp rise in fraudulent refund claims across China’s e-commerce platforms, with beauty emerging as the hardest-hit category.
What started as isolated manipulation in toys and apparel has evolved into a coordinated gray-market practice that now targets cosmetics. Beauty is particularly exposed because of its massive order volumes and thin merchant margins. One of the first brands to go public was Flower Knows, a Hangzhou-born color cosmetics label, which reported refund-without-return requests backed by AI-fabricated images.
As both shoppers and sellers learn to game refund systems with synthetic visuals, trust across platforms is eroding fast.
When AI shifts from efficiency tool to fraud amplifier
The surge did not begin in beauty. Toy sellers have seen AI-altered photos that make dolls look aged, cracked, or filthy. Apparel merchants receive images with exaggerated rips and stains. Homeware sellers are sent pictures of cups and bottles showing realistic-looking fractures that never existed.
Now the spillover into beauty is accelerating.
Ironically, the same AI tools that brands once embraced for content creation, virtual try-ons, and skin diagnostics are being repurposed to commit fraud. In one Flower Knows case, a customer submitted AI-generated images of a supposedly shattered blush compact to demand a refund. When the brand requested the item back, the customer reportedly smashed the intact product before returning it.
Similar schemes are spreading across the category. Merchants report fake “broken” false lash bands, digitally altered images of leaking foundation pumps, and palettes with synthetic scratches. Online tutorials advertising “30 successful refunds per account” are openly sold, dramatically lowering the barrier for fraud and encouraging copycat behavior.
The targeting is strategic. Most claims involve mid- to low-priced beauty products, usually under 100 RMB (about $14), where margins are slim and disputing a refund often costs more than the item itself. Small and mid-sized sellers are the most exposed.
A widening trust gap for sellers and shoppers
For merchants, especially SMEs, even contesting a 50 RMB ($7) claim can be economically irrational. Appeals require time, documentation, and technical expertise. In response, many sellers now record standardized packing videos, add traceable packaging IDs, and implement manual review steps, all of which drive up operating costs.
Consumers feel the impact too. As fraud rises across categories, platforms are imposing stricter proof requirements on legitimate buyers. These can include multi-angle videos, timestamps, and in beauty cases, even medical documentation for alleged skin reactions. After-sales service has become more adversarial, making genuine complaints harder to resolve.
A troubling countertrend is also emerging. Some merchants reportedly advise customers to upload AI-generated images to trigger platform safeguards, causing negative reviews to be flagged as malicious. This two-sided manipulation further undermines confidence in ratings and post-purchase support.
An AI governance gap for platforms and regulators
AI was meant to bring efficiency to China’s e-commerce supply chain. Instead, its misuse is hollowing out the trust that enables high-frequency digital shopping, from toys and T-shirts to cosmetics.
The core challenge is no longer adoption, but governance. Restoring confidence will require coordinated action: platforms strengthening detection systems, regulators tightening legal thresholds, merchants investing in responsible verification, and consumers acting in good faith. Only then can AI move back toward its intended role as a productivity engine rather than a fraud accelerator.
We have helped 20+ companies in industries like Finance, Transportation, Health, Tourism, Events, Education, Sports.
