What happened
On August 27, 2026, the Hong Kong Monetary Authority, Securities and Futures Commission, Insurance Authority, and Mandatory Provident Fund Schemes Authority — in collaboration with Cyberport — jointly announced the first cohort of the expanded Generative AI Sandbox++ (GenA.I. Sandbox++). The cohort selected 36 agentic-AI use cases from nearly 100 proposals, involving 30 financial institutions (including Manulife Hong Kong and Hang Seng Bank) and 27 technology partners. Unlike the prior GenA.I. Sandbox (content-generation focus), Sandbox++ specifically targets agentic AI systems performing or coordinating tasks with autonomy across end-to-end processes such as customer onboarding, payments, insurance claims, and anti-fraud. Participants will be onboarded to a Cyberport-managed AI Supercomputing Centre platform, with technical trials starting later in 2026.
Why it matters
This is a coordinated cross-sector supervisory initiative (banking, securities, insurance, pensions regulators) establishing a controlled, risk-managed testing environment specifically for agentic AI in financial services — an area regulators globally (MAS, RBI, APRA) are also moving into. While not a binding rule, it signals the supervisory framework and governance expectations (auditability, human oversight, controls) that HK regulators will expect before agentic AI reaches production in regulated financial institutions, and may generate de facto guidance/standards.
Action needed
Financial institutions operating in Hong Kong should monitor sandbox outcomes and emerging supervisory expectations on agent governance, auditability, and human oversight for agentic AI; non-participants should track published lessons-learned as a preview of future HKMA/SFC/IA supervisory guidance.