Hong Kong banks test AI deepfake detection in HKMA sandbox

Hong Kong banks reported results from AI deepfake detection trials under the HKMA's GenA.I. Sandbox. The second cohort includes 27 projects from 20 banks and 14 technology partners, with trials starting in early 2026. Ping An Digital Bank reported detection accuracy above 96%, while HSBC said none of roughly 20 deepfake attempts bypassed its controls.

• Ping An Digital Bank reports deepfake detection accuracy above 96%
• HSBC: none of ~20 test deepfake attempts bypassed its controls
• Second cohort: 27 projects from 20 banks and 14 partners
• HKMA sandbox trials begin in early 2026