Mr. Zhang, a fig tree grower from Xinji City, Hebei Province, opened a new store on a live streaming e-commerce platform last month with high hopes, selling his own cultivated fig saplings. However, just after selling 12 orders, he unexpectedly received a "partial refund" request, with the buyer claiming that the saplings had withered and died upon arrival.

The buyer attached a photo of a sapling with completely yellow leaves and no signs of life to support the refund request. Mr. Zhang, who has years of planting experience, immediately recognized the photo as a fake evidence created using AI technology.

Platform arbitration closes eyes and approves in minutes, merchants can't prove their innocence

Mr. Zhang said he always ships with the pot and ensures the soil is moist. The entire delivery process, from shipping to customer receipt, took only two days. According to the growth cycle of saplings, natural dehydration and wilting would take at least five to seven days, and the leaves would wrinkle and droop, not turn yellow so quickly as shown in the photo.

To protect his rights, Mr. Zhang promptly rejected the request and proactively negotiated for a return and refund. However, the buyer then requested platform intervention. To his disappointment, even though Mr. Zhang tried to explain with professional knowledge, the platform still ruled in favor of the buyer based on the AI-fabricated image, making a decision within just a few minutes.

National authorities introduce AI content verification, industry advocates using AI to combat AI

This 45-yuan "partial refund" order left Mr. Zhang feeling utterly wronged, causing him to lose all confidence in the platform. In frustration, he immediately initiated the process to close his store. This incident also revealed the new pain points of rampant AI-generated images and the difficulty in accurately identifying malicious online shopping claims.

In response to the increasingly rampant AI fraud, the National Anti-Telecom Fraud Center App has recently officially launched an AI content verification function, supporting the identification of images and videos. Industry security experts also pointed out that the future core solution lies in "using AI to fight AI," through the collaborative judgment of large models and small models to prevent such malicious free-riding phenomena at the source.