The Securities and Futures Commission (SFC) plans to develop principles-based guidance for AI applications with greater autonomy, focusing on safety, human accountability and oversight.

Chief Executive Officer Julia Leung outlined the plans at Hong Kong Fintech Week on Monday (2 November).

The regulator will draw on its Large Firm AI Readiness Survey and trials under GenAI Sandbox++ when developing further guidance.

SFC survey shows growing use of agentic AI

The survey covered 20 large financial groups representing about 75% of securities turnover and 45% of clients reported by licensed corporations.

All use AI for research, analytics and back-office support, while more than half are using or piloting agentic AI. About 80% expect further adoption over the next 12 months.

Leung warned that agentic AI could compound errors, take unauthorised actions and disrupt other firms. She said client advice, portfolio decisions and market-facing execution would require stronger safeguards than lower-risk workflows.

SFC plans cybersecurity framework

The SFC also plans a cybersecurity framework setting baseline requirements for all licensed firms, with enhanced measures for higher-risk firms.

Leung cited recent software vulnerabilities and said 90% of surveyed groups are using or evaluating frontier AI for cyber defence. Nearly half also use AI for adversarial simulation and threat hunting.

Tokenised retail products reach US$1.3 billion

By September 2026, the SFC had authorised 17 tokenised retail products with US$1.3 billion in assets under management, up 34% over 12 months.

Leung also highlighted Hong Kong’s decision to allow secondary trading of tokenised money market funds on licensed virtual asset trading platforms.

The HKMA also announced 24/7 interbank settlement in central bank digital currency to support round-the-clock tokenised transactions.

Featured image credit: Edited by Fintech News Hong Kong, based on image by 4045 via Magnific