Hong Kong's securities watchdog said it will provide additional guidance on agentic artificial intelligence adoption as appropriate, as a recent survey showed more than half of financial firms already use or are piloting it.
As human review for every intermediate step is unrealistic, the Securities and Futures Commission will be developing practical, principles-based guidance for applications that involve greater autonomy, with a focus on preserving human accountability and oversight throughout the workflow, said Julia Leung Fung-yee, Chief Executive of the SFC.
This came as agentic AI can compound errors across multiple steps, take unauthorised actions and disrupt other firms in unintended ways, she noted at Hong Kong FinTech Week 2026.
The SFC's latest survey also showed that 30 percent of investigated firms already use agentic AI in their business operations and 25 percent are piloting it. Conducted between July and August, the survey covered 20 large financial firms with 78 affiliated licensed corporations in Hong Kong.
Furthermore, the SFC plans to set out a cybersecurity framework with baseline requirements for all licensed firms, supplemented by enhanced requirements for firms with higher risk profiles, aiming to tackle AI-related cyber threats.
Citing recent global alerts of critical flaws in PHP and other web development tools, Leung said these flaws may allow attackers to take control of an unpatched server without logging in or accessing sensitive information.
On securities tokenization, Leung revealed that the SFC had authorised 17 tokenised retail products as of September, with their assets under management totalling US$1.3 billion (HK$10.14 billion), up 34 percent year-on-year.
