BOC Hong Kong and BOC Life Join GenA.I. Sandbox for Agentic AI Pilots

The Hong Kong Monetary Authority, the Securities and Futures Commission, the Insurance Authority, and the Mandatory Provident Fund Schemes Authority, together with Cyberport, have announced the list of participants for the first phase of the Generative Artificial Intelligence (GenA. I.) Sandbox++. Selected participants will be placed on a dedicated platform at Cyberport's Artificial Intelligence Supercomputing Centre, with technical testing scheduled to begin in late 2026.

The first phase of GenA. Sandbox++ will focus on exploring agentic artificial intelligence applications. Beyond content generation, the selected use cases will examine how to responsibly increase the autonomy of AI, with testing covering end-to-end processes in areas such as customer account opening, payments, insurance claims, and customer interaction. Building on the previous GenA. Sandbox theme of "AI fighting AI," the latest pilot projects will also assess how AI can be used to dynamically monitor the autonomous operations of other AI applications.

ICBC (Asia), which participated in the second phase of the GenA. Sandbox in 2025 and explored three scenarios, has again been selected as one of the participating banks in GenA. In line with regulatory guidance, it will conduct research and development, testing, and implementation of its selected scenario, the "Intelligent System for Compliance Risk Analysis of New Products." The system interprets current regulatory provisions, simulates professionals conducting comprehensive assessments across different risk categories, automatically identifies potential risk points, and carries out product information sorting, regulatory framework matching, and risk dimension breakdown to ensure new products meet regulatory requirements. Testing has already been completed on more than ten new products involving over 3,000 report items, with plans to expand the product and assessment scope.

AXA has been selected as one of the first participating institutions in GenA. It will use the computing resources of Cyberport's AI Supercomputing Centre to develop and test an application scenario focused on AI governance, cyber defense, and risk mitigation. AXA will explore advanced technical functions in a risk-controlled environment. Livi Bank has announced that its "AI vs. AI" anti-fraud intelligent prevention and control solution has been selected for GenA. Through a closed-loop process of mutual confrontation and mutual evolution, the solution enables AI to continuously learn from evolving financial crime methods, enhancing the bank's prevention and control capabilities and achieving self-evolution in fraud prevention.

BOC Hong Kong and BOC Life jointly announced their selection for GenA. Sandbox++, where they will conduct agentic AI proof-of-concept trials in two scenarios: BoC Pay+ consumer experience and life insurance frontline training. Building on the successful experience of the first two phases of the sandbox, the bank will run a six-month proof-of-concept, working with BOC Life and partners from different sectors to verify the feasibility and safety of agentic AI in consumer experience and frontline training.