Hong Kong’s financial regulators have introduced an expanded generative artificial intelligence sandbox as insurers globally increase AI investment and move deployments beyond pilot projects.
On March 5, the Hong Kong Monetary Authority (HKMA), Securities and Futures Commission (SFC), Insurance Authority (IA), and Mandatory Provident Fund Schemes Authority (MPFA) - in partnership with Hong Kong Cyberport Management Company Limited - launched the Generative Artificial Intelligence Sandbox++ initiative.
The new programme builds on the GenAI Sandbox launched in 2024. It now applies across banking, securities and capital markets, asset and wealth management, insurance, mandatory provident fund operations, and stored value facilities. The framework continues to emphasise three areas:
It also promotes “AI versus AI” approaches, using AI tools to help identify, monitor, and mitigate risks stemming from AI adoption. Participating institutions will receive supervisory guidance, technical support, and complimentary access to graphics processing unit (GPU) capacity at Cyberport’s AI Supercomputing Centre. Regulators say the arrangement is intended to provide a controlled setting for testing and refining AI models and workflows before they are scaled in live environments. Eddie Yue, chief executive of the HKMA, said the launch “marks a significant milestone under our ‘Fintech 2030’ strategy, reinforcing our commitment to building a vibrant ecosystem for responsible innovation.” By bringing together regulators, financial institutions, and the technology sector, the initiative aims “to unlock AI’s full potential to drive growth, efficiency, and customer-centricity across financial services,” he said.
The Sandbox++ supports both sector-specific and cross-sector applications. Examples cited by the regulators include AI-enabled underwriting and claims processing, tools to support suitability and compliance checks in investment product distribution, and AI-based solutions for MPF administration. Common applications across sectors include customer chatbots and fraud detection systems. Clement Cheung, chief executive officer of the IA, described the GenAI Sandbox++ as “a collective drive made by the financial regulators to nurture an environment conducive to AI innovation anchored upon accountability, inclusiveness, and prudency.” He said this thinking is consistent with the IA’s AI Cohort Programme, which has generated interest in “knowledge sharing and use case deployment,” and is helping attract talent that “should help to strength the competitiveness of Hong Kong as a regional AI hub.”
Read next: Are your staff actually any good at AI?
In the pensions space, MPFA managing director Cheng Yan-chee said the authority “strongly encourages the MPF trustees and intermediaries to actively explore the adoption of advanced financial technology solutions.” By using AI, he said, trustees and intermediaries can “enhance operational efficiency and elevate the quality of services, thus bringing more value to benefit MPF scheme members.” SFC chief executive officer Julia Leung said that widening participation in the sandbox among capital markets entities is “a true testament to our collective resolve to drive responsible market innovation and a tech-enabled and future-proof financial market.” She encouraged licensed corporations to take part to “harness the boundless potential of AI in enhancing operational efficiency, bolstering resilience, and unlocking growth.”
The Hong Kong initiative comes as new research indicates that insurers worldwide plan to raise AI investment in 2026, including in Asia-Pacific. Accenture’s “Pulse of Change” survey, conducted between November and December 2025, polled 3,650 C‑suite executives across 20 industries and 20 countries, including 218 senior leaders from insurance organisations. According to the findings, 90% of insurance executives intend to increase spending on AI in 2026. The research suggests most executives view AI primarily as a way to support revenue rather than as a cost-cutting measure. Eighty-five percent of the insurance leaders surveyed said they see greater benefits from AI for growth than for cost reduction.
The survey results indicate that insurers are beginning to integrate AI into multiple parts of their organisations. Accenture reported that 34% of insurance companies in the sample are actively deploying AI agents across several functions, rather than limiting the technology to isolated pilots. Generative AI is also entering daily routines for a subset of senior leaders. Almost one-third of insurance C‑suite respondents said they use generative AI tools every day, and 57% said they use them at least once a week.
In addition, 29% of surveyed organisations said they are redesigning end-to-end processes with AI as a core element of the workflow. This includes underwriting and pricing, claims triage and handling, distribution and servicing, and internal support functions. Survey respondents indicated that they would maintain or even step up their AI focus in the event of a market correction. If an AI “bubble” were to burst, 47% of insurance executives said they would increase AI investment and 37% said they would increase hiring.
The research also points to how insurance leaders view different types of disruption. Two-thirds of executives said they are prioritising investment in AI and digital technologies in response to accelerating change. About 67% reported feeling prepared for technology-related disruption, while smaller proportions said they feel prepared for environmental (39%) and geopolitical (44%) disruption. Despite expectations that 2026 will bring more change than 2025 – cited by 84% of respondents – most executives remain positive about business and workforce trends. Seventy-eight percent expect faster revenue growth this year, and 82% plan to hire more staff. The combination of Hong Kong’s Sandbox++ framework and rising global AI investment creates a context in which AI is likely to feature more in underwriting, claims, distribution, and operations. Market participants in the region are monitoring how supervisory initiatives, sandboxes, and access to computing resources influence the pace and direction of AI adoption in their businesses.
