Artificial Intelligence: Regulatory Landscape in China and Hong Kong

Introduction

Artificial intelligence (“AI”) is now a useful tool in many aspects of our lives. The public release of AI systems using generative pre-trained transformers in large language models has sparked widespread and popular adoption of GenAI tools. This, in turn, has attracted the attention of legislators and regulators. The goal for society is to maximise the benefits brought by AI, while mitigating risks. The mission for legislators and regulators is to find the right framework to achieve that goal.

Nevertheless, the regulatory approach to AI can differ from one jurisdiction to another. In August 2024, the European Union (“EU”) enacted its Artificial Intelligence Act, which was the first comprehensive risk-based AI regulatory framework that applies to all member states across the EU.[1] On the contrary, the United States does not have a unified AI regulation but adopts a case-by-case strategy with numerous guidelines to regulate AI on a federal level. In China, regulators have adopted more targeted and sector-specific rules on AI.

Hong Kong has adopted a context-based approach to regulation of AI, and has avoided the general legislative framework approach adopted by the EU. Under this approach, Hong Kong regulators have published regulations and guidelines within existing frameworks to provide guidance on AI in the context in which it is used. These sector and context specific regulations share common principles, but adopt an approach that can be more flexible to the needs of the sector. The requirements in the communications sector, for instance, will be different from the energy sector, which is in turn different to the financial services sector. Even within context-based regulation, there is a special role for the Privacy Commissioner for Personal Data (PCPD). This subject domain regulator has taken the lead in providing cross-sector guidance on the protection of personal data in the development, supply and use of AI systems in Hong Kong.

Hong Kong is a special administrative region of China. This means that although Hong Kong is a part of China, it has a separate system of law and regulation, including in relation to the governance of AI systems. In this article, we will focus on the AI legislative developments in Mainland China and Hong Kong.

What is Ernie Bot?

On 16 March 2023, Chinese tech giant Baidu officially launched its chatbot, Ernie Bot. It is one of the first generative AI chatbots to obtain regulatory approval in Mainland China. Similar to its main rival, ChatGPT, users can interact with Ernie Bot via text to create content. It primarily operates in Mandarin Chinese but can respond to English queries to a lesser extent. Although Ernie Bot is available for download globally, it requires a Chinese number for user registration. As such, the primary users of Ernie Bot are the Chinese population.  

In April 2024, Baidu announced that Ernie Bot had over 200 million users. The rising popularity of Ernie and other AI chatbots has resulted in the consideration of a more comprehensive legislative framework on AI to ensure that AI development, supply and use benefits China’s national interest.

[1] Link

AI regulation in Mainland China

In recent years, Mainland China has implemented the following major legislation and measures on the regulation of AI:

• Generative AI Measures: The Interim Measures for the Management of Generative Artificial Intelligence Services[2] (“Measures”) were jointly issued by the CAC and the National Development and Reform Commission, along with five other key government ministries on 10 July 2023, and took effect on 15 August 2023. The Measures are the first administrative regulation on the management of generative AI services in Mainland China, which apply to any individual that uses generative AI technology to provide services to the public in Mainland China.
• Regulation on algorithmic recommendations: Management of Algorithmic Recommendations in Internet Information Services Provisions[3] is Mainland China’s first legislation regulating the use of algorithm recommendation technologies to provide online services in Mainland China. The provisions came into effect on 1 March 2022. The purpose is to prohibit algorithmic recommendation service providers from generating fake news or disseminating information from unauthorised sources.
• Regulations on deep synthesis: Provisions on the Administration of Deep Synthesis Internet Information Services[4]came into force on 10 January 2023, following their adoption by the Cyberspace Administration of China, the Ministry of Industry and Information Technology, and the Ministry of Public Security on 25 November 2022. The regulations aim to address the risks associated with the AI-based “deep synthesis” (deepfake) technology which can generate or edit image, audio, or video content that appears to be authentic but is synthesised.
• Ethical principles and frameworks: The Measures for Review of Scientific and Technological Ethics (for Trial Implementation) came into effect on 1 December 2023. Under the legislation, enterprises engaged in AI and certain other technology activities are required to undergo scientific and technological ethics reviews.

The legislative developments above reflect that the Chinese authorities have adopted more targeted and sector-specific rules to regulate the evolving AI technologies. These regulations and measures do not apply in Hong Kong, as Hong Kong is a separate legal jurisdiction (although it is part of China).

[2] Full text of the legislation is only available in Chinese [Link]

[3] Full text of the legislation is only available in Chinese [Link]

[4] Full text of the legislation is only available in Chinese [Link]

AI regulation in Hong Kong

There is no overarching legislation regulating the use of AI in Hong Kong. The Hong Kong authorities have relied on existing legislation with sector-specific guidelines from regulators to address the risks and challenges posed by AI.

The PCPD first published general guidance on the topic of AI in its Guidance on the Ethical Development and Use of Artificial Intelligence (“Ethical Guidelines”) in 2021.[5] This paper was a general paper on AI governance frameworks, and adopted a broader outlook than just personal data protection. It positioned the PCPD as the natural general thought leader among regulators in Hong Kong on this topic.

On 11 June 2024, the PCPD issued the Artificial Intelligence: Model Personal Data Protection Framework[6] (“Model Framework”). The Model Framework is and is an elaboration of the Ethical Guidelines to the specific topic of personal data protection. While complying with the Personal Data (Privacy) Ordinance (Chapter 486 of the laws of Hong Kong)[7], the Model Framework provides a set of recommendations on the best practices for any organisations procuring, implementing, and using AI systems that involve the use of personal data.

Under the Model Framework, organisations are encouraged to formulate appropriate policies, practices and procedures when they procure, implement and use AI technologies:

AI Strategy and Governance: organisations are advised to have an internal AI governance strategy, which generally comprises an (i) AI strategy, (ii) governance considerations for procuring AI solutions, and (iii) an AI governance committee (or similar body) to steer the process.

Risk Assessment and Human Oversight: organisations are advised to adopt a risk-based approach in the procurement, use and management of AI systems. For instance, organisations should adopt a comprehensive risk assessment to systematically identify, analyse and evaluate the risks involved in the use of AI technologies.

Customisation of AI Models and Implementation and Management of AI Systems: the PCPD suggested that organisations should adopt rigorous testing and validation of the AI models in proportion to the level of risks involved, to ensure that the AI technologies will perform as intended.  

Communication and Engagement with Stakeholders: organisations are encouraged to communicate and engage effectively and regularly with stakeholders on the usage of AI.

In addition to this general guidance framework, different industry sector regulators have published guidance on different aspects of AI governance. For instance, the Hong Kong Monetary Authority is the regulator for banks in Hong Kong. It has published guidelines on AI generally[8], GenAI[9], and use of AI in transaction monitoring[10], and has opened a sandbox for the trialling of AI systems and applications under the sponsorship and input of the HKMA. Even professional sectors such as the Law Society of Hong Kong have provided guidance to its profession.[11]

[5] Link

[6] Link

[7] Link

[8] Link

[9] Link

[10] Link

[11] Link

What is next?

The approaches adopted in Mainland China and Hong Kong are a good example of the different approaches adopted around the world. Neither has adopted the general, unified regulatory model favoured by the EU. However, Mainland China is more likely to adopt additional legislative measures. The population of Mainland China is substantial, and the potential risks to social stability and welfare are correspondingly larger. Hong Kong has adopted, and is likely to continue, a context-based regulatory model that relies on existing legislation and regulatory frameworks.

This difference in approach between Mainland China and Hong Kong is also an excellent example of the continuing relevance and application of the one country, two systems policy of Mainland China that allows Hong Kong to operate a different legal system according to common law principles.

Ernie Bot, like many other GenAI systems, is available in Hong Kong. OpenAI’s ChatGPT is not (though workarounds are commonly applied). Nonetheless legislators and regulators in Mainland China and Hong Kong are alive to the benefits and challenges to society of GenAI technologies.

Pádraig Walsh and Stephanie Sy

If you want to know more about the content of this article, please contact:

Pádraig Walsh

Partner | Email

Disclaimer: This publication is general in nature and is not intended to constitute legal advice. You should seek professional advice before taking any action in relation to the matters dealt with in this publication. This article was last reviewed on 12 October 2024