Xiamen Firm Develops Deepfake Detection After AI Video Call Scam

A Chinese company in Xiamen has developed technology to detect AI-generated content, including deepfake videos and cloned voices, following a rise in scams where fraudsters impersonate family members via video calls to steal money. The firm helped draft a national standard requiring AI content to be labeled. The company, Meiya Pico, created the "Meiya Huiyan" platform that identifies nearly 500 types of forgery across images, videos, audio, and text. It provides probability scores and pinpoints tampered areas. A WeChat mini-program allows users to upload suspicious files for instant analysis, and has been integrated with anti-fraud platforms across 14 provinces. The technology is deployed in 27 provinces and regions, serving police, courts, banks, and universities. It operates offline for judicial cases to ensure data security. The company continues expanding its detection capabilities as AI forgery methods evolve, aiming to become a foundational safety infrastructure for the AI era.