Hong Kong’s financial regulators have launched GenA.I. Sandbox++, expanding a 2024 pilot into a cross-sector program that now covers banking, securities, insurance, MPF and stored value facilities. The initiative aims to speed up “responsible” GenAI adoption with supervisory guidance, technical support, and access to GPU computing resources.
The program is being led jointly by the HKMA, SFC, IA, and MPFA, in collaboration with Cyberport.
A cross-sector GenAI sandbox, not just banking
Sandbox++ broadens the original sandbox to multiple financial sectors, explicitly including securities and capital markets, asset and wealth management, insurance, mandatory provident fund services, and stored value facilities.
Regulators said the sandbox keeps its focus on three “high-impact” areas: risk management, anti-fraud, and customer experience, while also promoting “A.I. vs. A.I.” approaches—using AI to manage risks created by AI systems.
What participants get: guidance, support, and GPU time
Participating financial institutions will receive targeted supervisory guidance and technical support, plus complimentary access to GPU computing resources at Cyberport’s A.I. Supercomputing Centre. The goal is to let firms develop and refine GenAI use cases in a risk-controlled environment before scaling.
Example use cases regulators want to see
Sandbox++ encourages both sector-specific and cross-sector applications. Examples listed include:
- AI-driven insurance underwriting and claims processing
- Suitability-compliance assessment during investment product distribution
- AI tools for MPF operations
- Industry-wide tools such as intelligent customer chatbots and advanced fraud detection systems
Officials frame it as a “Fintech 2030” milestone
HKMA Chief Executive Eddie Yue called Sandbox++ a milestone under the regulator’s “Fintech 2030” strategy, positioning it as a coordinated push to improve efficiency and customer outcomes while keeping governance and risk controls in place. Leaders from the SFC, IA, and MPFA also emphasized responsible innovation, resilience, and the operational benefits of AI across their respective sectors.
Why it matters for crypto
- Stronger AI-driven fraud detection can raise the bar for on/off-ramps and reduce scam losses tied to crypto rails.
- Cross-sector AI pilots may speed up tokenized product distribution and compliance tooling (especially suitability checks).
- Institutional adoption of GenAI with regulator input can tighten risk and surveillance expectations for crypto-facing firms in Hong Kong.
- GPU-backed experimentation could accelerate real-time monitoring and AML-style analytics across digital asset flows.
What to watch next
- Which banks, brokers, insurers, and MPF operators publicly join the sandbox and what use cases they pick first.
- Whether regulators publish practical supervisory guidance or case studies from Sandbox++ deployments.
- Any cross-sector pilots that move from “customer service” into transaction monitoring and fraud interdiction at scale.
- Follow-on updates from Cyberport on capacity, onboarding, and how GPU access is allocated to participants.
Source: Hong Kong Monetary Authority