Hong Kong Introduces AI Guidelines to Protect Consumers in Finance

The Hong Kong Monetary Authority (HKMA) has introduced new principles aimed at governing the use of generative artificial intelligence (AI) in consumer-facing applications. These guidelines focus on enhancing governance, transparency, and data protection to safeguard consumers in an increasingly AI-driven financial landscape.

Strengthening AI Governance and Accountability

The HKMA, Hong Kong's central banking institution, has expanded its existing framework to include these new principles. Originally established in November 2019 for big data analytics and AI, the framework’s four pillars—governance and accountability, fairness, transparency and disclosure, and data privacy and protection—have now been updated to address the specific challenges posed by generative AI.

Alan Au, Executive Director of the HKMA’s banking conduct department, emphasized the rapid adoption of generative AI by banking institutions in recent months. Applications of this technology range from customer chatbots and personalized product development to targeted sales, marketing, and robo-advisors in wealth management and insurance. Au highlighted the importance of transparency, recommending that institutions provide clear and understandable disclosures to end-users.

Emphasis on Data Protection and Fairness

The HKMA’s new guidelines also place a strong emphasis on data protection. Authorized institutions are expected to ensure that generative AI models produce fair and consistent outcomes, and that customer data is rigorously safeguarded. The board and senior management of these institutions are held fully accountable for any decisions related to the deployment of generative AI.

Launch of the Generative AI Sandbox

On August 13, the HKMA, in collaboration with Cyberport—a state-run technology company—launched the “Generative Artificial Intelligence (GenAI) Sandbox” for the finance sector. This platform is designed to test AI’s potential applications in various financial services, including risk management, anti-fraud measures, customer service enhancements, and process reengineering.

Carmen Chu, Executive Director at the HKMA, explained that the GenAI Sandbox aims to help banks pilot innovative generative AI use cases within a risk-managed environment. The sandbox provides essential technical support and targeted supervisory feedback, overcoming barriers to AI adoption by offering both computing capabilities and regulatory guidance.

As generative AI continues to evolve, the HKMA’s proactive approach aims to ensure that its adoption in consumer-facing applications remains safe, transparent, and fair, ultimately protecting the interests of the end-users.