AI voice cloning used in financial scams across China and Thailand
Criminals in China and Thailand are using AI-powered voice cloning and synthetic media to conduct phone scams, including targeting of wealthy individuals in China and deceptive advertising in Thailand. In one reported incident in China, a victim lost 4.3 million yuan within 10 minutes to an AI voice scam.
What happened
Criminals in China are using AI voice cloning technology to conduct financial fraud, recording victims' voices from brief phone calls and then using AI models to replicate their voices convincingly enough that family members cannot distinguish the clones from the originals.
One victim reportedly lost 4.3 million yuan within 10 minutes to an AI voice impersonation scam, according to information disclosed by an official from Hangzhou Beichuang Information Technology Co., Ltd. China's Ministry of Public Security Cyber Security Bureau issued anti-fraud warnings in May 2026 about the increasing prevalence of voice cloning fraud.
In Thailand, scammers are using virtual videos and voice manipulation software to impersonate real people, creating fake single woman personas to target middle-aged men, using tactics described as the 'Ben Lao Tou' scam, involving small repeated deceptive payments rather than large single sums.
Reported impact
Victims in China have suffered direct financial losses through AI voice cloning scams. The reported case of 4.3 million yuan lost in 10 minutes demonstrates the significant economic harm possible.
In Thailand, the 'Ben Lao Tou' scam uses repeated small payments that accumulate; while individual amounts may be small (such as the cost of tea or food delivery), multiple victims across dozens of targets can result in theft exceeding 3,000 yuan, which constitutes criminal fraud under Chinese law.
Index analysis
The incident reflects vulnerabilities in voice authentication systems and phone-based communication verification processes. Criminals can capture sufficient voice data from very brief interactions—as little as 5 seconds—to train AI models capable of producing highly realistic synthetic speech.
The scams target both high-value victims (in the 4.3 million yuan incident) and large numbers of lower-income targets through accumulation of small payments, exploiting psychological factors such as social embarrassment and perceived futility of reporting small-value crimes.
Governance controls
China's National Anti-Fraud Center released preventive measures including recommendations that people minimize verbal responses to unknown callers, avoid posting original audio and video content on social media platforms, establish verification protocols with family members and business associates, and use a new AI authentication verification feature in the National Anti-Fraud Center app to detect synthetic media.
Classification
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.