Hong Kong pauses proposed AI copyright amendment for guidelines

Hong Kong will delay legislation on copyright protection for works used to train artificial-intelligence models and will first develop operational guidance, according to the South China Morning Post. Officials and lawmakers said the pause reflected unresolved concerns over enforcement, licensing and the balance between creative rights and AI development.
Hong Kong authorities have postponed a proposed amendment on the use of copyrighted works to train artificial-intelligence models and will first prepare operational guidelines, the South China Morning Post reported. The Copyright (Amendment) Bill 2025 was intended to introduce an opt-out mechanism for text and data mining. It had been scheduled for submission to the Legislative Council but was removed from its legislative agenda last August.
The proposal was not included in the city’s 2026 policy blueprint or its first five-year plan, according to the report. Critics had questioned how copyright holders would tell AI companies not to use their works and had suggested an opt-in system instead. Lawmaker Johnny Ng Kit-chong said the government would first draft guidance on how the opt-out mechanism could operate.
He said copyright holders seeking royalties would need to present a workable system for implementation. Commerce and Economic Development Secretary Algernon Yau Ying-wah said the government had received differing views during a two-month consultation in 2024. He said officials needed to consider the concerns and develop a better solution through changes, recommendations or guidelines.
The proposed bill included a text-and-data-mining exception allowing reasonable use of copyrighted works for computational analysis and AI training. Proposed safeguards included lawful access to material and requirements to maintain and disclose records of sources. Lawmakers and industry representatives welcomed a cautious approach while raising concerns about compensation, licensing and enforcement.
Other issues mentioned included AI agents’ potential legal liability, the difficulty of tracing copyright owners across jurisdictions and the possible use of encrypted watermarks.
This independently written report is based on information supplied by the named publisher. Vertrix News has not independently verified the source report.