---
title: "AI face and voice clone steals $622K in China video scam"
slug: "ai-face-and-voice-clone-steals-622k-in-china-video-scam"
published: "2026-09-23"
beat: "Crime"
tags: ["Crime"]
creator: "Agentry Newsroom"
editor: "Susanne Sperling, Editor — Human in the Loop"
tools: ["Claude (Anthropic)", "Perplexity Sonar"]
creativeWorkStatus: "verified"
dateReviewed: "2026-09-23"
aiActArticle50: "compliant"
humanView: "https://agentry.news/crime/ai-face-and-voice-clone-steals-622k-in-china-video-scam"
agentView: "https://agentry.news/agent/ai-face-and-voice-clone-steals-622k-in-china-video-scam"
---# AI face and voice clone steals $622K in China video scam

> Police in Baotou, Inner Mongolia reported a scammer used AI-generated facial and voice clones to impersonate a victim's friend on a video call, stealing approximately $622,000 in September 2026. Most 

*Drafted by an AI agent. Verified by Susanne Sperling, Editor — Human in the Loop. [AI policy](/ai-policy).*

A scammer in Baotou, Inner Mongolia used AI-generated facial and voice clones to impersonate a victim's friend during a video call, stealing approximately $622,000 before most of the funds were recovered [Agentry News](https://agentry.news/agent/ai-face-voice-clone-steals-622k-in-china-video-scam).

## Attack Method and Victim Impact

The fraud relied on synthetic media — AI-cloned face and voice — to establish false trust during a real-time video interaction. The attacker impersonated someone known to the victim, a social engineering tactic amplified by generative AI's ability to replicate biometric markers in real time. The victim transferred $622,000 before the deception was discovered [AI Policy Tracker](https://aipolicytracker.org/ai-risk/incidents/992).

The case exemplifies a growing category of **agent-enabled fraud** where autonomous or semi-autonomous AI systems execute the social engineering and transaction steps at scale. Voice cloning and deepfake video can now run in real-time video applications without requiring hours of rendering — enabling live impersonation that traditional synthetic media attacks could not achieve.

## Recovery and Regulatory Context

Local police successfully recovered most of the stolen funds, though the full recovery amount was not disclosed in available reports. The incident surfaced amid China's tightening regulatory framework on AI-generated content. China's Supreme Court has begun establishing legal precedent for deepfake liability [StartupFortune](https://startupfortune.com/chinas-supreme-court-sets-first-legal-rules-for-ai-deepfakes-and-lies/), and deepfake voice cloning is now subject to liability rules under emerging Chinese AI governance [Cryptopolitan](https://www.cryptopolitan.com/china-liability-rules-for-ai-deepfakes/).

The Baotou case is one of the first documented high-value video-call fraud incidents using real-time face and voice synthesis. It reflects the closing gap between synthetic media capability and criminal deployment — a gap that had existed when deepfake tools required studio-grade rendering. Live cloning tools are now accessible enough to appear in street-level fraud operations, not just nation-state espionage.

## Broader Implications for Agent Security

The incident raises questions about identity verification in agentic systems. If AI agents conduct transactions or communications on behalf of humans, or verify identities via video, live deepfake cloning becomes a vector for transaction fraud at scale. Unlike traditional chatbot impersonation, video-call deepfakes exploit the last layer of trust — visual and audio confirmation.

No court sentencing or regulatory penalty was disclosed in available sources. Police in Baotou have not released the suspect's name or a prosecution timeline.