Skip to content
Submit incident
AII-1001-39453
Well documentedThree or more independent publishers report the core facts.Supporting sources are pending verification.

AI-generated models and livestreams mislead online clothing shoppers in China

Chinese e-commerce merchants increasingly use AI-generated models and livestream hosts to display clothing without prominent disclosure, creating significant mismatches between advertised and actual products in color, fabric, fit, and design details. Service platforms have made AI content generation highly efficient and low-cost, reducing production expenses by over 90% while customer service responses avoid confirming AI use.

9 October 2025 Β· Chinese e-commerce merchants Β· Misinformation and synthetic media
9 October 2025
AII-1001-39453
Quick facts
Category
Misinformation and synthetic media
Country
China
Date
9 October 2025
Drafted by
Raidu Incident Detection

What happened

Chinese e-commerce merchants use AI-generated models and clothing images in product listings and livestreams without clear disclosure, causing significant visual misrepresentation between advertised and actual products.

AI model images are generated using uploaded product photos paired with template models from service platforms like LinkFox AI, which can produce thousands of variants daily with computer-generated details such as embroidery that do not exist on actual items.

AI livestream hosts, created either through capturing real people's movements in studios or using template-based digital personas with text-to-speech synthesis, conduct 24/7 product demonstrations. These hosts use preset keyword response systems and scripted dialogue to simulate real-time interaction.

Reported impact

Consumers report severe discrepancies between AI-generated images and received products, including incorrect colors, cheap synthetic fabrics instead of advertised materials, poor fitting, and non-existent design details.

One consumer purchased a dress after viewing AI-generated images that appeared 'like a magazine spread', but received an item with 'huge color difference', 'cheap chemical fiber', improper fitting, and embroidery details that did not exist on the actual product.

Another consumer purchased a suit after watching an AI host demonstrate tailored fit and motion, but received garments with soft, unsupportive fabric that could not hold the shape, realizing only afterward that even the host's walking and turning movements were AI-generated.

Index analysis

A clothing e-commerce operator estimated that adopting AI technology reduced visual content production costs by over 90%, as AI can generate thousands of product images per day compared to dozens using real models, while also switching between different ethnic backgrounds and locations instantly.

The technology has created a production industry chain where service providers supply AI image generation and livestream host creation, with merchants requiring only a flat product image and a base model image to generate complete product displays.

Current e-commerce customer service responses avoid directly confirming whether images are AI-generated, instead stating that 'all images are real shots' while using 'slight color difference between product and image may exist' as legal cover, indicating industry-wide evasion of disclosure.

Governance controls

A lawyer stated that AI products and virtual hosts cannot bear legal responsibility themselves, as they function as tools or mediums; actual responsibility lies with the organizations operating them, including brand operators, MCN agencies, or livestream marketing personnel, with technology service providers potentially bearing joint liability if they produce illegal content.

China's 'Artificial Intelligence Generated and Synthesized Content Labeling Measures' effective 1 September 2025 requires that images, videos, virtual scenes and other content generated by AI technology must carry prominent labels identifying them as AI-generated.

E-commerce platforms are required to establish and enforce labeling rules as a precondition for content review and apply effective constraints to violators, though monitoring bodies have not yet provided clear definitions of what constitutes an AI livestream or who is responsible for monitoring.

Classification

Category
Misinformation and synthetic media
Industry
Logistics, wholesale and retail
Country
China
Harm status
Actual harm
Intent
Intentional
Timing
Unknown
Harm types
Economic or property
Harmed
Consumers
Failure modes
Deepfake image, Deepfake video, Manipulation or dark pattern
AI systems
AI livestream hosts, AI model generation systems, Artificial voice synthesis
Regulations
GDPR (cited)

Sources and evidence

Sources not yet published

This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.

How sources are verifiedSuggest a source for this incident