Abstract:
"Although model pictures were heavily Photoshopped in the past, at least I could see the fabric and texture. Now I use AI directly. Do I want to sell it to AI?" Many netizens complained on social platforms that the e-commerce AI model pictures were "distorted" and "unable to judge the pattern." As Double Eleven is approaching, the chaotic display of e-commerce AI models has triggered heated discussions. An investigation by The Paper found that in some stores, AI models had twisted hands and three legs. There were even positive photos of AI models with "three legs" in the comment section. The flaw is so obvious, but the AI-generated logo is almost invisible, making it difficult to tell whether it is genuine or not, leaving consumers confused.
The picture of a girl model in an online store, the hand posture obviously violates the human body structure.
Behind the chaos is a mature AI model industry chain. To take a real photo of a piece of clothing, the model, venue, and shooting are packaged for at least two to three hundred yuan; the AI-generated model picture costs only 6 yuan per piece, and the cost advantage is clear at a glance. On the Internet, a collection of AI prompts for women's clothing pictures and tutorials on the workflow of bringing goods are sold at low prices, with tens of thousands of copies sold. If a merchant sends you a picture of a garment, you can get the AI-produced film in half an hour, but the actual shooting often takes a day or two. On the one hand, there are practical demands of merchants to reduce costs and increase efficiency, and on the other hand, consumers continue to complain that the goods are not as shown in the pictures. Some platforms have successively introduced rules to rectify fake AI pictures and clarify AI labeling requirements.
First of all, it should be acknowledged that it is not illegal for businesses to use AI model pictures. Using AI to generate virtual models for product display is a normal application of artificial intelligence and is not prohibited by current laws. Technology itself has no original sin. Even real-life pictures may be over-edited and divorced from the real thing. AI pictures are also legitimate as long as they faithfully restore the product. As many netizens said: "Real-person displays may not be completely realistic. I do not reject well-done AI."
AI is not the original sin, but concealment is a trap. What the law restricts is never technical tools, but misleading and fraud. Whether a merchant’s use of AI models constitutes false publicity or consumer fraud, the criterion is never “whether AI is used”, but whether the displayed images are consistent with the real thing, and whether the information is deliberately concealed. As long as the fabrics and styles are objectively presented, no fabrication of quality is made, no flaws are concealed, AI attributes are actively marked, and no false inducements are made, the red line of consumer fraud will not be touched.
On the other hand, if there is a huge gap between the AI renderings and the actual product, breaking through the boundaries of reasonable beautification, and adding customer service lies and prevarication, deliberately erasing the AI logo, etc., it will be easily identified as consumer fraud. According to Article 55 of the Consumer Rights Protection Law, merchants must bear punitive damages of “refund one and compensate three”. AI has never been a gold medal for businesses to avoid liability, nor is it an invisibility cloak for fraud. Anyone who tries to use AI as a shield will have to pay for false propaganda.
The "Measures for the Labeling of Synthetic Content Generated by Artificial Intelligence" has long made it clear that clear and perceptible labels should be set for AI-generated graphics, text and videos; concealment without labeling is enough to mislead consumers, which is a violation of the right to know. The platform also has the obligation to verify. For content that detects traces of synthetic traces but does not declare it, it should promptly mark the content and act as a "gatekeeper" for consumer trust.
AI models "capture" Double Eleven, so there is no need to stop eating due to choking. Tools are neither good nor evil; the responsibility lies with the person who uses them. Only by using AI properly and keeping the bottom line of not defrauding consumers can AI models get out of the vortex of public opinion where authenticity is difficult to distinguish, shed the negative label of "counterfeiting tools", truly take root in the real economy, serve consumers, and become a credible and easy-to-use digital helper in the e-commerce industry.
Comments