When China began opening its economy in 1978, Hong Kong offered a window onto international markets. Its manufacturers expanded production across the border, while its trading networks and professional services helped connect mainland businesses with overseas customers and capital.
Hong Kong’s role in financial intermediation was strengthened after its return to China in 1997. The city subsequently deepened its connections with mainland markets through Stock Connect in 2014 and 2016, followed by Bond Connect in 2017. These arrangements connected investors across markets while accommodating different rules and institutions.
Sustainable finance has extended Hong Kong’s intermediation role even further. The Common Ground Taxonomy, developed through an EU–China working group under the International Platform on Sustainable Finance, compared their green classifications. Hong Kong drew on this work in developing its own taxonomy in 2024, mapping activities against mainland and European classifications.
These experiences illustrate institutional intermediation: helping participants navigate different systems and identify where their requirements overlap. They provide a foundation for Hong Kong’s next contribution in the artificial intelligence (AI) age.
China’s strengths in clean-technology manufacturing and its position among the world’s leading AI developers create opportunities to combine intelligent systems with renewable energy and efficient infrastructure to advance sustainable development. Hong Kong could connect these capabilities with international finance and markets.
Hong Kong’s first five-year plan, published in September, gives this ambition a policy foundation. It proposes AI and data governance aligned with national and international rules. The practical challenge is to develop arrangements that participants from different systems can trust and use.
Yet AI adoption also creates dependencies. Reliance on a small group of technology providers can limit financial institutions’ choices, while common models can narrow the range of judgments informing their decisions.
The Financial Stability Board has warned that common AI models and data could increase correlations in lending, trading and pricing. This raises a concern: institutions may each retain human oversight while the financial system collectively loses diversity of judgment. The FSB’s sound governance practices for AI adoption emphasise leadership responsibility, human oversight and the management of provider dependencies.
Hong Kong has begun addressing AI governance in banking. On October 15, the HKMA announced its ART framework, centred on accountability, resilience and trust. It includes updated supervisory guidance and further work on AI agents that can plan and execute tasks. Its effectiveness will depend on implementation. As systems gain autonomy, oversight must include the practical ability to intervene in their actions and assign responsibility when things go wrong.
There is already discussion of a broader role for Hong Kong: it has been proposed that Hong Kong could serve as a testing ground for AI governance across jurisdictions.
AI governance raises questions of social justice. Who benefits from greater productivity, who bears the costs of disruption, and how do people secure a livelihood? In an analysis, the IMF warns that AI could increase inequality and reduce labour income, while emphasising the importance of enabling adoption and supporting workers.
Hong Kong could demonstrate how effective governance enables technological progress while advancing social justice. This means enabling rapid adoption alongside credible means of making a living for people whose jobs disappear. Although the employment effects remain uncertain, further advances in AI need not eliminate people’s role in production.
The five-year plan’s proposed upgrade of the Employees Retraining Board into Upskill Hong Kong offers a starting point. Government and business could jointly develop training linked to employment opportunities, assistance with self-employment and income support, including longer-term support where necessary. Corporations benefiting most from AI-driven productivity gains should contribute fairly to financing this transition, and employee participation should help shape transition arrangements without becoming a general veto over productive technology.
Becoming a model for inclusion would be an achievement Hong Kong could earn through deliberate effort by government, businesses and society.
Success should be judged by whether people can adapt to changing work or find new employment, whether small and medium-sized enterprises can use AI productively, and whether those unable to secure sufficient earnings receive adequate support. Higher aggregate output alone would not demonstrate an inclusive transition.
Hong Kong’s distinctive institutional position provides a foundation for such an ambition. Under “one country, two systems”, Hong Kong has distinct governing arrangements alongside its common-law system and financial institutions. Its experience connecting mainland and international markets could help participants understand different expectations about authority and accountability.
This could give practical expression to what I have called “participation without alignment”. Organisations cooperating through Hong Kong could agree on specific tasks without requiring their countries to share a common political system or an exclusive geopolitical alignment. Confidence would depend on how transparently and reliably particular arrangements operate.
For example, participants from China, Europe and Southeast Asia could develop shared tests for specific AI applications. They would need agreement on evidence, responsibility and dispute procedures, while respecting applicable laws. Smaller firms and economies should have practical opportunities to participate. Cooperation would have to accommodate remaining differences rather than assume universal recognition of a single standard. Shared testing could reduce duplicated assessments, allowing participants to devote more resources to deployment while retaining the safeguards required in their own markets.
Hong Kong’s value would be demonstrated when cooperation brings useful technologies into operation, preserves accountable judgment and helps people share in greater prosperity. If Hong Kong can turn these ambitions into demonstrable results, it could offer a practical example of how different political, legal and value systems can cooperate on shared problems without first resolving their wider differences.
Shouqing Zhu is Senior Policy Advisor at UNEP Finance Initiative.
