AI deepfake scams target elderly and financial systems in China
Criminals in China are using AI deepfake technology to commit fraud, targeting elderly people with voice-cloned impersonations and financial systems with synthetic video attacks. Online services offer voice cloning and face-swapping at low cost, while financial companies like 度小满 have deployed detection systems to identify and block such attempts.
What happened
Criminals are using AI deepfake technology to commit fraud in China, including facial deepfakes and voice cloning. Scams target elderly people by impersonating their relatives and requesting emergency money transfers, and target financial systems through fake liveness verification videos.
Voice cloning and face-swapping services are available online at low prices, from 1 yuan to around 100 yuan, with 'one-stop services' offered by dark industry operators on overseas platforms.
Reported impact
In one case, an elderly woman in Hunan (Liu Nainai) was deceived by a voice-cloned call and transferred 30,000 yuan after believing her grandson needed emergency assistance.
Fraudsters target the elderly by exploiting their unfamiliarity with new technology, limited information channels, and emotional dependence on family. Scams also include impersonating doctors and experts to sell products with exaggerated health claims.
Index analysis
Deepfake detection systems show that AI-generated faces now achieve over 90% realism in some cases. A risk scoring system identifies deepfakes: scores above 50 points indicate high risk, with synthetic faces typically scoring above 50 and authentic faces below 50.
Governance controls
度小满 has developed specialized risk identification algorithms and upgraded its liveness detection models to billion-level parameters, compared to tens of millions previously. The company uses a 'red-blue adversarial' training approach simulating criminal AI attacks to prepare defenses.
In 2024, 度小满 issued fraud warnings to 140,000 customers and successfully blocked over 3,000 fraud cases, reportedly preventing approximately 180 million yuan in economic losses. Industry experts recommend cross-sector information and technology sharing to prevent repeat attacks.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.