Hong Kong to issue AI agent safety guidelines for society-wide adoption in 2027

Hong Kong plans to introduce AI agent safety management guidelines in 2027 as part of a wider effort to strengthen governance of artificial intelligence (AI) development and deployment.

Professor Sun Dong, the Secretary for Innovation, Technology and Industry, announced the plans on November 4 at the 2026 World Internet Conference (WIC) Wuzhen Summit in Zhejiang province, China.

Speaking at a forum on interoperability and mutual trust among intelligent agents, Sun said the Hong Kong Special Administrative Region (HKSAR) Government would jointly develop the guidelines with the Hong Kong Artificial Intelligence Research and Development Institute. The proposed guidelines will apply across society, rather than being limited to government departments.

The initiative forms part of Hong Kong’s broader approach to AI development, which focuses on expanding the adoption of AI and AI agents, strengthening governance frameworks, and promoting international exchange and cooperation.

Sun said the government would also support the national Global AI Governance Initiative and promote international collaboration through large-scale AI conferences in Hong Kong.

The announcement follows a keynote address delivered by Sun at the WIC ministerial meeting on November 3, in which he outlined Hong Kong’s seven-area AI risk governance strategy. The areas include combating AI-related crime, strengthening protections for minors, assessing ethical risks associated with AI, promoting safe adoption across industries, examining legal liability for AI-related accidents, issuing guidelines for AI agents, and assessing the impact of AI on employment.

The strategy is intended to balance the promotion of AI innovation with measures to manage potential risks, while maintaining a people-centred approach to development and deployment.

Hong Kong is also expanding its AI infrastructure and investment. Sun said the government’s computing capacity had reached 5,000 petaflops, while the Sandy Ridge Data Facility Cluster, currently under construction, is expected to provide 180,000 petaflops by 2032.

The government has also established a Committee on AI+ and Industry Development Strategy to coordinate the integration of AI across industries, involving government, industry, academia, research and investment stakeholders. An additional HK$1 billion is being allocated to the Artificial Intelligence Funding Scheme to support research and the development of intelligent computing industries.

The proposed AI agent guidelines come as governments and regulators increasingly examine the governance challenges associated with autonomous AI systems, including their ability to perform tasks, interact with other systems and act on behalf of users.

However, the Hong Kong government has not yet outlined the specific requirements or enforcement arrangements that the forthcoming guidelines will introduce. Further details are expected as work on the framework progresses.