
AI rules for Hong Kong businesses
The AI laws and guidelines that matter in Hong Kong, Mainland China, the EU and the US, and the dates that are coming up.
General information, not legal advice. Check the official source before you act.
Key dates ahead
Hover over or tap a date to see what changes.
2026
2 December 2026
EU AI ActNew bans on AI that generates non-consensual intimate imagery or child sexual abuse material apply, and generative AI systems placed on the market before 2 August 2026 must meet the Article 50(2) machine-readable marking duty.
9 December 2026
New Product Liability Directive (PLD)Member States must transpose the Directive by this date, and the new liability rules apply to products placed on the market or put into service after it.
31 December 2026
2026 Policy Address AI governance planThe FSTB is expected to hold a public consultation later in 2026 on a legal framework for detecting and removing fraudulent content, including AI-generated fraud.
31 December 2026
2026 Policy Address AI governance planThe Hong Kong AI Research and Development Institute (AIRDI) is expected to begin operation in 2026.
2027
1 January 2027
CCPA ADMT regulationsBusinesses already using ADMT for significant decisions must comply with the ADMT rules on pre-use notice, opt-out and access.
1 January 2027
California AI Transparency Act (content provenance)Duties for large online platforms and GenAI hosting platforms begin, requiring them to detect and display provenance data.
1 January 2027
California GenAI developer and chatbot lawsSB 867's ban on toys containing companion chatbots takes effect.
1 January 2027
Colorado ADMT law (formerly Colorado AI Act)SB 26-189 takes effect, and the Attorney General's rules are due by this date.
12 January 2027
Data ActProviders of cloud and other data processing services may no longer charge switching fees.
20 January 2027
Machinery RegulationThe Machinery Regulation applies and the Machinery Directive 2006/42/EC is repealed, with no transition period.
31 January 2027
PIPLThis is the annual deadline for reporting the 2026 compliance audit on minors' personal information.
1 March 2027
DOJ bulk sensitive data rule (Data Security Program)Annual reports covering calendar year 2026 are due for restricted cloud-computing transactions with 25% or more country-of-concern or covered-person ownership.
13 March 2027
US outbound investment rules (OISP / COINS Act)This is the 450-day statutory deadline for the Treasury to issue regulations implementing the COINS Act, and existing rules apply until then.
1 July 2027
California GenAI developer and chatbot lawsSB 243 annual reports to the Office of Suicide Prevention begin, and SB 1119 core duties are expected to start.
1 July 2027
US AI-in-hiring lawsCalifornia SB 947 becomes operative.
2 August 2027
GPAI model obligations and Code of PracticeGeneral-purpose AI models placed on the market before 2 August 2025 must comply.
12 September 2027
Data ActUnfair contract term rules extend to certain long-term data-sharing contracts concluded on or before 12 September 2025.
2 December 2027
EU AI ActObligations for high-risk AI systems listed in Annex III apply, for example in recruitment, credit scoring, education, biometrics and critical infrastructure.
11 December 2027
Cyber Resilience Act (CRA) and RED cybersecurity rulesThe CRA applies in full, covering essential cybersecurity requirements, conformity assessment and CE marking, and the RED cybersecurity Delegated Regulation (EU) 2022/30 is repealed.
31 December 2027
2026 Policy Address AI governance planAI agent safety-management guidelines for society at large, prepared with the AIRDI, are expected in 2027.
31 December 2027
CCPA ADMT regulationsRisk assessments must be completed for processing that began before 2026 and continues.
2028
1 January 2028
California AI Transparency Act (content provenance)Capture device manufacturer duties apply to cameras, phones and recorders first produced for sale in California from this date.
1 April 2028
CCPA ADMT regulationsFirst risk assessment submissions to CalPrivacy are due, along with the first cybersecurity audit certifications for businesses with 2026 revenue above US$100 million.
2 August 2028
EU AI ActObligations for high-risk AI embedded in regulated products covered by Annex I apply.
2 August 2028
Machinery RegulationThis is the legal deadline for Commission delegated acts adding AI Act-based requirements for high-risk AI safety functions to the Machinery Regulation.
2029
1 January 2029
Connected vehicles rule (ICTS)The ban on importing vehicle connectivity system (VCS) hardware with a China or Russia nexus applies to components without a model year, and from Model Year 2030 to others.
1 April 2029
CCPA ADMT regulationsCybersecurity audit certifications are due for businesses with revenue of US$50–100 million.
2030
1 January 2030
Colorado ADMT law (formerly Colorado AI Act)The 60-day right to cure before enforcement expires.
1 April 2030
CCPA ADMT regulationsCybersecurity audit certifications are due for smaller covered businesses.
1 August 2030
General Product Safety Regulation (GPSR)In a related sector rule, the new Toy Safety Regulation (EU) 2025/2509 applies, including rules for connected and AI-enabled toys.
2031
1 January 2031
California GenAI developer and chatbot lawsSB 867's toy ban expires unless extended.
Find a rule
12 rules found
Showing the Hong Kong rules. Choose other markets or search to see more.
Hong Kong 12 rules
Hong Kong governs AI mainly through the PDPO and guidance from the Privacy Commissioner, the Digital Policy Office and the financial regulators, with further plans in the 2026 Policy Address.
- AI laws and governanceGuidanceGuidanceHigh relevance
PCPD AI Model Framework
Artificial Intelligence: Model Personal Data Protection Framework (with the supplementary guidance 'Protecting Personal Data Privacy in the Use of Agentic AI')
The Privacy Commissioner's main reference for organisations that buy, customise or use AI systems involving personal data. It recommends measures in four areas, covering AI strategy and governance, risk assessment and human oversight, customisation and management of AI systems, and stakeholder communication. A 2026 supplement applies the same approach to agentic AI. It is guidance, not law.
- Issuer:
- Office of the Privacy Commissioner for Personal Data (PCPD)
- Last verified:
- 1 October 2026
- Official sources:
- Artificial Intelligence: Model Personal Data Protection Framework (PCPD, PDF), PCPD media statement, 11 June 2024, Protecting Personal Data Privacy in the Use of Agentic AI (PCPD, PDF), PCPD media statement, 25 August 2026 (agentic AI guidance), PCPD media statement, 19 May 2026 (AI compliance checks on 60 organisations), PCPD media statement, 16 March 2026 (agentic AI alert), Guidance on the Ethical Development and Use of Artificial Intelligence, 2021 (PCPD, PDF), PCPD AI Privacy Protection resource page
Questions about what this means for your business? Ask the secretariat
- Workplace and hiringGuidanceGuidanceHigh relevance
PCPD GenAI Employee Checklist
Checklist on Guidelines for the Use of Generative AI by Employees
A practical checklist to help organisations write an internal policy on staff use of generative AI tools at work while complying with the PDPO. It covers permitted tools and purposes, what data may be entered, checking outputs, security and the consequences of breaches. It is guidance, not law.
- Issuer:
- Office of the Privacy Commissioner for Personal Data (PCPD)
- Last verified:
- 1 October 2026
Questions about what this means for your business? Ask the secretariat
- AI laws and governanceGuidanceGuidanceHigh relevance
DPO GenAI Guideline
Hong Kong Generative Artificial Intelligence Technical and Application Guideline (read with the Ethical Artificial Intelligence Framework)
The Government's practical guide for developers, service providers and users of generative AI in Hong Kong. It sets out a governance framework built on five dimensions (personal data privacy, intellectual property, crime prevention, reliability and trustworthiness, and system security) and recommends that AI-generated content be identified by watermarks, labels or metadata. The companion Ethical AI Framework provides principles and an AI assessment template. Both are voluntary.
- Issuer:
- Digital Policy Office (DPO), HKSAR Government; Guideline prepared with the Hong Kong Generative AI Research and Development Center (HKGAI)
- Last verified:
- 1 October 2026
- Official sources:
- DPO: Ethical AI Framework and Generative AI Guideline page, Hong Kong Generative AI Technical and Application Guideline v1.1 (DPO, PDF), Ethical AI Framework v2.0 (DPO, PDF), DPO press release, 15 April 2025, LCQ11: Regulating use of AI-generated synthetic content (15 July 2026), 2026 Policy Address, paragraph 123(iii) (procurement declarations)
Questions about what this means for your business? Ask the secretariat
- Personal data and privacyIn forceLawHigh relevance
PDPO
Personal Data (Privacy) Ordinance
Hong Kong's general data protection law. It is technology-neutral and principle-based, so its six Data Protection Principles apply whenever an AI system collects, uses, stores or produces personal data. Hong Kong has no AI-specific statute, which makes the PDPO the main binding law for AI that touches personal data.
- Issuer:
- Legislative Council of the HKSAR; regulated by the Office of the Privacy Commissioner for Personal Data (PCPD)
- Last verified:
- 1 October 2026
- Official sources:
- Personal Data (Privacy) Ordinance, Cap. 486 (Hong Kong e-Legislation), PCPD: The Ordinance at a Glance (principles and penalties), LCQ2: Prevention of personal data breaches and financial crimes (22 January 2025), PCPD paper to LegCo Panel on Constitutional Affairs (13 February 2026), CMAB paper to LegCo Panel on Constitutional Affairs (23 September 2026)
Questions about what this means for your business? Ask the secretariat
- Personal data and privacyIn forceMeasuresHigh relevance
GBA Standard Contract (Hong Kong)
Standard Contract for Cross-boundary Flow of Personal Information Within the Guangdong-Hong Kong-Macao Greater Bay Area (Mainland, Hong Kong)
A voluntary, pre-set contract that lets organisations move personal information between Hong Kong and the nine Mainland cities of the Greater Bay Area through a simpler process than the general Mainland export routes. Both parties sign the standard terms, carry out a simplified impact assessment and file the contract within 10 working days. This entry covers the Hong Kong side's requirements and benefits; the Mainland side is covered separately.
- Issuer:
- Cyberspace Administration of China (CAC) and HKSAR Innovation, Technology and Industry Bureau (ITIB); administered in Hong Kong by the Digital Policy Office; PCPD guidance
- Last verified:
- 1 October 2026
- Official sources:
- Digital Policy Office: GBA Standard Contract facilitation measure, Digital Policy Office: Facilitating cross-boundary data flow within the GBA, Implementation Guidelines on the GBA Standard Contract (English translation, PCPD, PDF), PCPD Guidance on Cross-boundary Data Transfer: GBA Standard Contract (PDF), Government press release: facilitation measure extended to all sectors (1 November 2024), PCPD: LegCo Panel on Constitutional Affairs meeting, 23 September 2026
Questions about what this means for your business? Ask the secretariat
- AI laws and governanceProposedPolicyHigh relevance
2026 Policy Address AI governance plan
The Chief Executive's 2026 Policy Address: Application and Risk Governance of Artificial Intelligence (paragraphs 102–123)
Hong Kong's first broad plan for governing AI risks. It does not create a general AI law, but it sets seven priorities, covering AI-enabled crime, protection of minors, an ethics governance framework, application safety, liability for accidents caused by AI products, AI Agent guidelines and the impact on employment. Most items are studies, consultations or guidelines, and none is yet law.
- Issuer:
- Chief Executive of the HKSAR; coordinated by the Chief Secretary for Administration, with a new Commissioner for AI under the Digital Policy Office
- Next key date:
- 31 December 2026. An FSTB public consultation on a legal framework for detecting and removing fraudulent content, including AI-generated fraud, is expected later in 2026.
- Last verified:
- 1 October 2026
- Official sources:
- Full text of the Chief Executive's 2026 Policy Address (4), paragraphs 102-123, 2026 Policy Address (full text, PDF), LCQ11: Regulating use of AI-generated synthetic content (15 July 2026), Digital Policy Office: Hong Kong AI Research and Development Institute, CMAB paper to LegCo Panel on Constitutional Affairs (23 September 2026)
Questions about what this means for your business? Ask the secretariat
- Financial servicesGuidanceGuidance
SFC GenAI circular
Circular to licensed corporations: Use of generative AI language models
The SFC expects licensed corporations that use generative AI language models in regulated activities to apply four core principles, namely senior management oversight, AI model risk management, cybersecurity and data risk management, and third-party provider risk management. Investment advice, recommendations and research are treated as high-risk uses needing extra safeguards. A June 2026 circular adds expectations on defending against AI-enabled cyberattacks.
- Issuer:
- Securities and Futures Commission (SFC)
- Last verified:
- 1 October 2026
Questions about what this means for your business? Ask the secretariat
- Financial servicesGuidanceGuidance
HKMA GenAI guidance
Consumer Protection in respect of Use of Generative Artificial Intelligence (HKMA circular) and related HKMA AI supervisory guidance
The HKMA expects banks using generative AI in customer-facing services to apply its 2019 big data and AI principles plus extra safeguards covering board accountability, fair outcomes, disclosure to customers and data protection. In the early stage of deployment, banks should keep a human in the loop and, as far as practicable, let customers opt out and ask for human review. These are supervisory expectations, not statute.
- Issuer:
- Hong Kong Monetary Authority (HKMA)
- Last verified:
- 1 October 2026
- Official sources:
- HKMA circular, 19 August 2024 (Banking Regulatory Document Repository, PDF), HKMA press release: GenA.I. Sandbox++ launched (5 March 2026), HKMA press release: First cohort of GenA.I. Sandbox++ (27 August 2026), HKMA press release: Fintech 2030 (3 November 2025), HKMA report: Supporting A.I. Adoption in Fighting Financial Crime (June 2026, PDF), HKMA circular: Use of AI for Monitoring of Suspicious Activities (9 September 2024, PDF)
Questions about what this means for your business? Ask the secretariat
- Generative AI and contentIn forceLaw
Deepfake intimate images offence
Crimes Ordinance offence of publishing or threatening to publish intimate images without consent (covers AI-altered and deepfake images), and the pending review of deepfake law
Publishing, or threatening to publish, an intimate image of a person without consent is an offence under the Crimes Ordinance, including where the image has been altered with deepfake technology. Creating such an image without publishing it is not a specific offence. The Law Reform Commission is studying deepfake-related crime and legislation is under consideration, but no bill exists.
- Issuer:
- Legislative Council of the HKSAR; enforced by the Hong Kong Police Force; law reform review by the Law Reform Commission and the Department of Justice
- Last verified:
- 1 October 2026
- Official sources:
- LCQ2: Combating use of AI technology to create indecent images (28 January 2026), LCQ11: Regulating use of AI-generated synthetic content (15 July 2026), 2026 Policy Address, paragraph 123(i), PCPD: Abuse of AI Deepfakes, Toolkit for Schools and Parents (2025, PDF; cites ss.159AA and 159AAE), Crimes Ordinance, Cap. 200 (Hong Kong e-Legislation)
Questions about what this means for your business? Ask the secretariat
- Generative AI and contentProposedLaw
Copyright TDM exception (proposed)
Proposed text and data mining exception for AI under the Copyright Ordinance
In 2025 the Government proposed a new copyright exception allowing copies of works to be made for computational data analysis, including AI model training, for commercial and non-commercial use, subject to conditions and an opt-out for rights holders. The bill was not introduced, and reports in September 2026 say the Government will first issue operational guidelines. The exception is not law.
- Issuer:
- Commerce and Economic Development Bureau (CEDB) and Intellectual Property Department (IPD)
- Last verified:
- 1 October 2026
Questions about what this means for your business? Ask the secretariat
- Products, consumers and cybersecurityIn forceLaw
Critical infrastructure cybersecurity law
Protection of Critical Infrastructures (Computer Systems) Ordinance
Hong Kong's cybersecurity law for critical infrastructure has been in force since 1 January 2026. Operators designated by the Commissioner must keep a security management unit, run risk assessments and audits, have an emergency response plan and report security incidents within tight deadlines. It is not AI-specific, but it covers AI systems that form part of designated critical computer systems.
- Issuer:
- Security Bureau; enforced by the Commissioner of Critical Infrastructure (Computer-system Security), with the HKMA and the Communications Authority as designated authorities for their sectors
- Last verified:
- 1 October 2026
- Official sources:
- Government press release: Ordinance to come into effect on 1 January 2026 (27 June 2025), Government press release: Appointment of Commissioner (1 January 2026), Communications Authority: PCICSO overview (obligations and Code of Practice), Office of the Commissioner of Critical Infrastructure (Computer-system Security): press releases, Cap. 653 (Hong Kong e-Legislation)
Questions about what this means for your business? Ask the secretariat
General information, not legal advice. This overview summarises public sources as last verified on 1 October 2026. Rules and dates change, and whether a rule applies to you depends on your circumstances. Please check the official source and take legal advice before you act on anything here.
Need help with AI compliance?
Several HKAIIA member companies work on AI governance, legal and compliance. Members can find them through member matchmaking and ask the secretariat for an introduction.