Deepfake videos used to impersonate financial analysts in investment fraud schemes
Fraudsters in China used AI deepfake videos impersonating financial analysts Li Xunlei, Zhang Xia, and Xun Yugeng to trick retail investors into signing fake investment agreements. At least one investor reportedly lost 3 million yuan. Multiple brokerages discovered the scams and issued fraud warnings in early 2026.
What happened
Fraudsters used AI technology to create deepfake videos impersonating Li Xunlei, chief economist at CITIC International, to deceive retail investors. The fake videos featured Li discussing how 'most retail investors get wiped out' in the U.S. stock market, with fraudsters then tricking investors into signing fake agreements for 'institutional investment accounts' promising guaranteed returns.
One investor reportedly deposited 3 million yuan into a fake account, only to find the account could not be opened and required Li Xunlei's password to access. Since early 2026, multiple head brokerages including Zhaoshan Securities, Guoxin Securities, and CITIC Securities discovered that fraudsters had impersonated multiple top analysts: Zhang Xia, chief strategist at Zhaoshan Securities, and Xun Yugeng, chief economist at Guoxin Securities.
Reported impact
At least one investor lost 3 million yuan to the fraud. The incidents show that fraudsters have obtained personal information including names, identification photos, and securities industry credentials from multiple analysts to carry out the schemes.
Index analysis
Deepfake technology was specifically deployed to manipulate video content, making it appear that the financial analysts had endorsed fraudulent investment schemes. The sophistication of the synthetic media allowed fraudsters to create convincing but entirely fabricated recommendations.
Governance controls
Multiple brokerages issued statements warning about the fraudulent schemes and advised investors to report incidents to police. No evidence indicates the financial institutions had implemented AI detection systems to identify or block the deepfake content before it was used in fraud schemes.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.