Deepfake fraud targeting celebrities and executives in China
Deepfake technology is being used to commit fraud against celebrities, business executives, and ordinary citizens in China. AI face-swap and voice-cloning tools are widely available on commercial platforms for under 10 yuan, enabling scammers to impersonate authority figures and deceive victims into transferring large sums of money.
What happened
Model 叶凯薇 was targeted with AI deepfake pornographic videos. AI face-swap software is sold on Chinese e-commerce platforms for as little as 2.88 to 50 yuan per use or as a subscription package for 30 yuan per month offering unlimited uses.
A Hong Kong engineering company employee received a video conference call with deepfaked video of the company's CFO and other executives discussing a secret acquisition. The employee believed the meeting was real and transferred 200 million Hong Kong dollars to five accounts as instructed, only discovering the fraud five days later when contacting the real headquarters.
Entrepreneur 雷军 became a victim of AI voice cloning and deepfake videos that circulated widely during China's October National Day holiday in 2024. Criminals have used AI to defraud individuals and businesses, including a case where a construction materials manager transferred 200,000 yuan after receiving a deepfaked video call from an alleged supervisor, and a Tianjin businessman defrauded of 950,000 yuan by an AI-synthesized leader video.
Reported impact
Celebrities, business executives, and ordinary citizens reported becoming targets of deepfake fraud. The technology enables criminals to impersonate authority figures and bypass visual verification systems.
Users can obtain deepfake services through social platforms, secondhand marketplaces, and e-commerce sites, with customized deepfake videos available for as little as 9.9 yuan when providing a five-second video and clear facial photos. Professional open-source software like DeepFaceLab combined with widely available tutorials has made the technology accessible to the general public.
Index analysis
Existing detection methods remain ineffective as deepfake technology produces results with high fidelity. However, the generated content still exhibits detectable artifacts such as missing micro-expressions, mismatched lighting, overly smooth skin, unnatural voice pauses, and fast speech patterns.
Platforms have struggled to balance AI content moderation with user engagement, adopting inconsistent policies. Tiktok and other content platforms require creators to label AI-generated content but face challenges in detection and enforcement. Media platforms have sometimes been effective at debunking AI-based rumors, as when Weibo CEO responded to false celebrity gambling rumors in March 2025.
Governance controls
An AI-generated content labeling guideline was proposed to require visible markers at generation and distribution stages to alert users to AI-generated content. Some e-commerce platforms reject requests for celebrity deepfakes citing portrait rights concerns, while others do not. Video platforms restrict uploading of illegal content through automated warnings.
Legal frameworks exist prohibiting the use of information technology to forge images and protecting dynamic video content, but lack quantified standards for damages and responsibility boundaries. Punishment remains disproportionate to profits gained, creating incentive for continued criminal activity. Experts recommend adding watermarks at the tool source level, strengthening platform detection, and enhancing legal regulations to clarify AI usage scenarios and legal consequences.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.