Opening Speech by the Acting Commissioner for Digital Policy, Mr Daniel Cheung, JP, at the "Fanovation 2026" (with photos)

Miles (Co‑Founder & CEO, Fano Labs, Dr Miles Wen), distinguished guests, ladies and gentlemen,

Good morning. It is a great pleasure to join you at Fanovation 2026. AI agents mark a fundamental shift in technology; moving from systems that merely generate text to ones that can execute work autonomously. This conference, focused on trusted agentic AI, is most timely in addressing the reality confronting every organisation.

Some of us may have already experimented with AI agents, or at least we understand what they can achieve. Today, the real operational challenge may be governance, predictability and control, such as defining where autonomy ends, preventing unintended actions, and establishing clear accountability. Trust is no longer an optional feature; it is an absolute prerequisite for deployment. Before any autonomous system is given discretion, enterprises and the public must be assured that safety boundaries, accountability and human oversight are firmly in place.

The need for robust AI governance aligns closely with the Government’s vision in the First Five‑Year Plan for Economic and Social Development (2026‑2030) and the 2026 Policy Address. Hong Kong remains committed to advancing the “AI+” Initiative, driven by the development directions of “industries for AI” and “AI for industries” strategy.

To steer these directions, the Government has set up the Committee on AI+ and Industry Development Strategy, chaired by the Financial Secretary, to create favourable conditions for expediting AI development and adoption. Through this high‑level coordination, we are accelerating deep AI integration across key sectors, including finance, healthcare, while supporting AI upgrades for small and medium enterprises. At the same time, we are enhancing the AI ecosystem in multiple dimensions, such as computing infrastructure, research and development, and talent cultivation.

Realising the full potential of these technologies demands that we treat operational risks with the same dedication we give to opportunities. The Government’s policy is technology for good underpinned by safety and control. Accordingly, we will be advancing our AI risk‑governance strategy in seven aspects, including combating criminal use of AI, promoting ethical risk assessments, and introducing guidelines for AI agents. This strategy is backed by the creation of the post of Commissioner for AI under my Office to assist various bureaux with the work.

The Government is also leading by example. The DPO is piloting AI agents internally and has embedded relevant safeguards in the AI Adoption Guide for Government Bureaux and Departments. Selected pilot departments will require bidders to comply with our AI guidelines and submit safety and ethics compliance declarations for AI technologies used in procuring public-facing AI service projects. Beyond these, the DPO will actively engage industry bodies to develop sector‑specific AI application guidelines, accreditation and governance frameworks, and to provide professional advice and support.

To broaden these efforts, the Hong Kong Artificial Intelligence Research and Development Institute (AIRDI), which will commence operation later this year, will bridge upstream R&D with industry needs and advise on AI governance. Drawing on our practical experience, we will work with AIRDI to publish AI Agent safety‑management guidelines for society at large next year.

Building a trusted AI ecosystem is a shared endeavour. While sound public policy provides the guardrails, true operational reliability comes from the dedication and discipline of organisations like yours that deploy and supervise these systems every day. I encourage you to share your insights openly so that together we can turn autonomous technologies into secure, scalable solutions that inspire lasting confidence.

I wish the conference great success and look forward to fruitful exchanges today. Thank you very much.

- ENDS -