HKMA combines current high-level AI and consumer-protection principles with supervised experimentation through an expanded GenAI sandbox.
Banks are expected to address governance, fairness, transparency, privacy and GenAI-specific risks such as hallucination and customer understanding.
Hong Kong authorised institutions, with specific expectations for customer-facing GenAI applications.
Summarised from the primary instruments below. Open the source before relying on an obligation for legal or compliance decisions.
Protect customers using AI-supported services and provide safeguards for customer-facing GenAI outputs.
Apply fairness, transparency and disclosure principles, including clear information about AI interactions and decisions.
Address hallucination, output reliability and human oversight in customer-facing generative applications.
The original authority remains the source of truth. Dates below are publication and effective dates, not OneBench ingest dates.
Hong Kong Monetary Authority
Extends existing AI consumer-protection principles to customer-facing GenAI and adds safeguards for new risks.
Hong Kong Monetary Authority
Expands supervised testing arrangements for GenAI use cases in authorised institutions.