TLDR: The Hong Kong Monetary Authority (HKMA) has launched the second phase of its Generative AI (GenA.I.) Sandbox, involving 20 banks and 14 technology firms. This initiative will test 27 new use cases focused on responsible AI deployment, enhanced deepfake fraud detection, risk management, and improving customer experience within the financial services sector. The trials are anticipated to commence in early 2026, building on the success of the initial cohort.
Hong Kong is taking a significant stride in integrating artificial intelligence into its financial landscape, as the Hong Kong Monetary Authority (HKMA) announced the selection of a second cohort of participants for its Generative AI (GenA.I.) Sandbox. This expanded initiative will see 20 banks and 14 technology firms collaborate to test 27 innovative use cases, with a strong emphasis on responsible AI governance and bolstering defenses against emerging threats like deepfake fraud.
The GenA.I. Sandbox, a joint effort between the HKMA and Hong Kong Cyberport Management, provides a controlled environment for financial institutions to develop and rigorously test AI-driven solutions. The second phase represents a substantial increase in participation and scope compared to the inaugural cohort, which involved 10 banks and four technology partners testing 15 use cases.
Arthur Yuen Kwok-hang, the HKMA’s Deputy Chief Executive, highlighted the importance of this expansion, stating, ‘The second cohort represents a major leap forward in making AI adoption safer and more robust.’ He further emphasized the industry-wide consensus on AI’s transformative potential, noting, ‘The diverse representation of banks in this cohort highlights the industry-wide consensus on AI’s transformative potential.’ Yuen also underscored the HKMA’s commitment to fostering the local fintech ecosystem through combined efforts from banks, technology firms, and academic institutions.
A defining theme for the 27 use cases in this new cohort is a proactive approach to AI governance, aiming to improve accuracy and consistency at a scalable level. Crucially, the sandbox will serve as a testing ground for advanced defense mechanisms against deepfake-related fraud. Several participants are slated to conduct adversarial simulations, utilizing AI to stress-test and fortify their systems against sophisticated attacks.
Beyond fraud detection, the second cohort will also focus on enhancing overall risk management and improving customer experience. A key addition to this phase is the introduction of the GenA.I. Sandbox Collaboratory. This platform will host a series of practical workshops designed to foster early engagement and collaboration between banks and technology providers, facilitating the conversion of problem statements into actionable AI-driven solutions. A dedicated Collaboratory workshop will specifically address combatting deepfake attacks with AI, bringing together experts to share solutions and explore innovative detection tools.
Also Read:
- Thailand’s Financial Sector Navigates AI Transformation Amidst High Expectations and Regulatory Scrutiny
- Generative AI Propels Sophisticated Academic Fraud in Chinese Paper Mills
Banks are encouraged to explore integrating AI into their second and third lines of defense for risk management, addressing both the opportunities and challenges presented by AI advancements. The trials for this second cohort are expected to commence in early 2026, following the application period which closed on August 31, 2025, and the official launch announcement on April 28, 2025.
