HKMA Launches ART Framework to Promote Responsible AI Adoption in Banking

By AI123

On October 15, the Hong Kong Monetary Authority (HKMA) unveiled the ART framework—Accountability, Resilience, and Trust—to support the responsible adoption of artificial intelligence in the banking sector. The framework aims to build an AI-enabled banking ecosystem grounded in these three pillars.

Noting that AI systems are becoming more autonomous and capable of mobilizing more data, tools, and workflows, the HKMA introduced several concrete measures. These include issuing an updated supervisory guidance on responsible AI application, which incorporates insights from the second phase of the GenAI sandbox; collaborating with Cyberport and the University of Hong Kong to build an AI training data sharing platform with data quality assurance as a core design principle; and expanding the GenA.I. Sandbox++ to further test agentic AI applications. The authority also published a discussion paper on a proposed regulatory framework for agentic AI governance, exploring intention-based accountability arrangements, a register of significant AI agents, and effective intervention mechanisms.

The HKMA also co-hosted the Finance + A.I. forum with Cyberport on October 15 and 16, marking the successful conclusion of the second-phase GenAI sandbox. More than 60% of the use cases explored agentic AI, reflecting the industry's active push toward advanced applications such as automatic planning and executing multi-step tasks within authorized boundaries.

The HKMA's Deputy Chief Executive said that the shift from generative AI to agentic AI marks an important new phase in banking innovation. Under the Fintech 2030 strategy, the ART framework provides a clear basis for banks to develop and deploy AI responsibly, combining innovation, clear accountability, resilient operations, and effective human oversight to fully realize AI benefits while maintaining trust and financial stability in Hong Kong.

Explore topicsAgentic AIAIHKMAPolicy

Sources

View all