On 19 Aug 2024, the HKMA issued guiding principles requiring authorized institutions to extend their existing 2019 BDAI Consumer Protection Framework to GenAI applications, mandating specific safeguards including human-in-the-loop oversight for early-stage customer-facing use, opt-out options for customers, transparent disclosure of GenAI limitations, and adherence to PCPD data privacy guidance. The HKMA also encourages leveraging GenAI to proactively enhance consumer protection through targeted vulnerability identification and fraud prevention.
This article was generated using SAMS, an AI technology by Timothy Loh LLP.
SFC Regulation of Virtual Asset Trading Platforms: The Emerging Hong Kong Framework
SFC proposes new regulation of virtual assets through licensing of operators of cryptocurrency and other digital asset trading platforms
Market Misconduct Tribunal: A Guide to Penalties and Other Consequences for Misuse of Insider Information
Market Misconduct Tribunal proceedings for mishandling of insider information can result in fines and disgorgement orders
SPOTLIGHT ON
FAMILY OFFICES
Introduction
On 19 Aug 2024, the Hong Kong Monetary Authority (HKMA) issued a circular providing guiding principles for authorized institutions on consumer protection in customer-facing generative artificial intelligence (GenAI) applications, extending the existing 2019 Big Data Analytics and Artificial Intelligence (BDAI) Guiding Principles to address GenAI-specific risks.
Extension of Existing Framework
The HKMA confirms that GenAI, as a subset of BDAI, shares similar risk dimensions but introduces heightened concerns including lack of explainability and hallucination risks. Consequently, the HKMA requires authorized institutions to apply and extend the 2019 BDAI Guiding Principles to GenAI use in customer-facing applications, maintaining a risk-based approach while implementing additional safeguards.
GenAI-Specific Safeguards
Under governance and accountability, institutions must define clear scopes for customer-facing GenAI, establish policies with control measures, and implement a 'human-in-the-loop' approach during early deployment to ensure accuracy. For fairness, institutions must prevent biased outcomes through data anonymization, comprehensive datasets, and provide customers with an opt-out option for GenAI-driven decisions, alongside channels for human review where opt-out is impracticable. Transparency requires clear disclosure of GenAI use, purpose, and limitations to customers. Data privacy mandates compliance with the Personal Data (Privacy) Ordinance and PCPD guidance, including the 2024 'Artificial Intelligence: Model Personal Data Protection Framework'.
Proactive Consumer Protection Opportunities
The HKMA encourages authorized institutions to explore GenAI for enhancing consumer protection, such as identifying vulnerable customers requiring additional safeguards, providing tailored product disclosures, and issuing fraud alerts for high-risk transactions, while ensuring these applications adhere to the outlined principles.
View the full article:Source
