Published • loading... • Updated
Ping An Digital Bank Completes HKMA’s Second GenA.I. Sandbox Cohort | Taiwan News | Oct. 15, 2026 19:20
Sandbox testing showed over 96% detection accuracy for deepfake fraud, the bank said, as it explored an AI-versus-AI defense strategy.
- Ping An Digital Bank Limited completed the Anti-Fraud Strategy Platform use case as part of the second cohort of the Hong Kong Monetary Authority's Generative Artificial Intelligence Sandbox on October 15, partnering with Hong Kong Cyberport Management Company Limited.
- Developed with OneConnect Financial Technology, the platform leverages Ping An Group's proprietary technology to combat deepfake-related fraud by exploring an 'A.I. vs A.I.' defense strategy that flags manipulated images in real time.
- Sandbox testing results demonstrated a detection accuracy of over 96% when identifying forged or synthetic faces, validating Generative Artificial Intelligence's potential to strengthen anti-fraud efforts and customer protection across the sector.
- Chief Executive Ronald Iu noted the sandbox served as an invaluable testing ground, while the project advanced anti-fraud development by enabling banks to share practical experience with industry peers through the HKMA process.
- Ping An Digital Bank plans to increase investment in Artificial Intelligence, deepening applications in risk control and product innovation while striving to become the "user-friendly digital bank" for every customer.
Insights by Ground AI
12 Articles
12 Articles
Ping An Digital Bank Tests Deepfake Detection in HKMA Sandbox
Ping An Digital Bank has completed its use case in the second cohort of the Hong Kong Monetary Authority’s (HKMA) GenAI Sandbox. According to the bank, its fraud platform detected deepfakes with over 96% accuracy in sandbox tests. The HKMA runs the sandbox together with Cyberport. It gives banks a space to test and explore [...]
Coverage Details
Total News Sources12
Leaning Left1Leaning Right1Center1Last UpdatedBias Distribution34% Left, 33% Center, 33% Right
Bias Distribution
- 34% of the sources lean Left, 33% of the sources are Center, 33% of the sources lean Right
34% Left
L 34%
C 33%
R 33%
Factuality
To view factuality data please Upgrade to Premium
