AI Governance in Asia: Navigating the Future of Technology
With the speed at which Artificial Intelligence (AI) is rapidly transforming various sectors across the globe, countries are left perplexed as to how to approach its regulation and governance. Some challenges surrounding its regulation include defining AI and ensuring the regulation does not hinder innovation and business activity. This article provides an overview of the AI regulatory landscape in Asian countries based on recent developments.
Singapore
Aiming to be at the forefront of AI regulation in Southeast Asia, Singapore approaches regulations by balancing business competitiveness and certainty within the law. It does so by opting for frameworks instead of establishing a universalised body of legislation.
Singapore: Direct regulations
Singapore started its journey through the establishment of the ‘Advisory Council on the Ethical Use of AI and Data’ in 2018. The Council consists of members appointed by the Minister of Communications and Technology. It was established to advise the government on ethical AI policy and its administrative implications on society and the greater nation. Furthermore, it aims to help the government guide businesses in using AI in their operations ethically and efficiently.
Aiding the Council’s efforts, the Infocomm Media Development Authority released ‘The Model AI Governance Framework’ in 2019. This framework guides organisations on how they can responsibly implement AI systems and development. This includes explaining how AI systems work, their ethical concerns, good data accountability practices and AI solutions. As of August 2024, the framework is in the process of being updated to address newer issues concerning AI for greater relevancy. This is to be finalised by the end of 2024.
The Infocomm Media Development Authority has further developed another tool, ‘AI Verify’ which seeks to help organisations validate their AI systems to ensure compliance against ethical principles. Some of these principles are, but are not limited to, security, human agency and oversight and accountability. After being tested against said principles, a report is produced to increase transparency among shareholders and the general public. Within ‘AI Verify’ is another initiative, ‘AI Verify Foundation (AIVF)’, a non-profit foundation which pools in experts from the private sector. These experts are in charge of developing AI testing frameworks, standards and practices to ensure Singapore’s AI governance is as relevant and up-to-date as possible.
Singapore: Indirect regulations
Besides a general framework surrounding the governance of AI, Singapore also mandates AI systems through indirect and more sector-specific frameworks. The Health Products Act 2007 mandates that any medical device incorporating AI must be registered before it can be used. This ensures that AI applications in healthcare meet stringent safety and efficacy standards before reaching the market. In the financial sector, the Principles to Promote Fairness, Ethics, Accountability, and Transparency (FEAT) lay down foundational principles for integrating AI in decision-making processes. These principles emphasize fairness and transparency, ensuring that AI-driven financial products are designed and deployed with ethical considerations at their core. Looking ahead, Singapore’s Personal Data Protection Commission has established a new division in 2024 focused on the use of personal data in AI recommendation systems.
Hong Kong
Similar to Singapore, Hong Kong does not regulate AI through mandatory or legally binding measures. Instead, they signal their expectations through the ‘Model Framework’ (“the Framework”) to privacy regulators and organisations of which they are recommended to follow upon an occurrence of data breach resulting from the use of of AI tools. Hong Kong: The Model Framework Being a major step towards AI governance in Hong Kong, Framework provides a set of non-exhaustive guidelines organisations should adopt within their AI systems. The Framework includes a set of Data Stewardship Values; being respectful, beneficial and fair and, Ethical Principles which are, but not limited to, accountability, transparency and data privacy. The Framework is segmented to cover different aspects of AI regulations. The first one includes AI Strategy and Governance, including a list of various strategy and governance considerations.
The second segment includes a Risk Assessment and Human Oversight Board which carries out risk assessments to weigh out and tailor the risk of an AI system to the organisation using it. The Board considers allowable uses of data, volume and sensitivity of data used, the quality of data, security of personal data and the probability of privacy risks arising from the system when conducting their risk assessment. However, this system needs to be careful not to be too rigid in its assessment as it may disproportionally weigh out the risks against very necessary AI output.
The third segment is the ‘Customisation of AI Models and Implementation and Management of AI Systems’. This area seeks to prepare and manage the data used in AI systems to ensure data and system security. It lays out a 3-step procedure. Firstly, data preparation. This is done by data minimisation which ensures each individual’s data is protected and data quality which ensures the AI output is fair and unbiased. Secondly, customisation and implementation. This confirms that the AI solutions meet the requirements of implementing transparency, traceability and auditability . Thirdly, management and monitoring which carries out risk assessments, periodic reviews and an overall evaluation of the AI landscape.
The fourth segment includes ‘Communication and Engagement with Stakeholders’. This ensures stakeholders are receiving information regarding how credible and transparent an AI system is. The information given includes how personal data is being used to customise and train AI solutions and how AI systems can impact individuals.
Thailand
Thailand is actively working on regulating and promoting AI through its strategic plans and regulations. However, in contrast to Singapore and Hong Kong, it chooses to adopt legislative measures.
Thailand: Legislative measures
The first legislative measure is the ‘Draft AI Royal Decree’ that adopts a risk-based approach. The Decree categorises AI systems based on their risk levels; prohibited, high-risk and limited risk. Each category includes a set of requirements the AI system must comply with. Prohibited systems include those that influence human behaviour subliminally and those that perform real-time remote biometric identification in public areas. In addition, The Decree requires all AI providers operating in Thailand to appoint a local representative to ensure compliance with the registration requirements.
The second measure is the ‘Draft Act on the Promotion and Support of AI Innovations’. This Act seeks to support AI initiatives such as AI sandboxes which facilitate the testing and innovation of AI. Moreover, it encourages data exchange for AI training to enhance the market competition within Thailand.
Thailand: Strategic plans and regulations
Thailand has a ‘National AI Strategy and Action Plan (2022-2027)’ which sets out a set of strategies and milestones Thailand aims to achieve an effective AI ecosystem by 2027.
This strategy includes:
- Legal and Ethical Readiness: Preparing regulations and ethical guidelines for AI application;
- Infrastructure Development: Investing in digital infrastructure to support AI;
- Education and Talent Development: Increasing AI literacy and creating AI professionals;
- Technology and Innovation: Promoting R&D and innovation in AI;
- Adoption in Public and Private Sectors: Encouraging widespread AI use in various sectors.
The strategy aims to achive the following milestones by 2027:
- Educating 600,000 people about AI laws and ethics;
- Enforcing comprehensive AI laws and regulations;
- Elevating Thailand's AI readiness to be among the top 50 globally;
- Investing 10% more annually in digital infrastructure;
- Developing over 30,000 AI professionals and 100 R&D prototypes;
- Generating significant economic and social impacts through AI innovations.
Looking Ahead
The AI regulatory landscape in Asia is evolving, with countries like Singapore and Thailand leading the way with comprehensive frameworks and strategic plans. These regulations aim to strike a balance between fostering innovation and ensuring ethical, fair, and secure AI usage. As AI continues to advance, ongoing updates and refinements to these regulations will be crucial in addressing new challenges and opportunities.
Bibliography
- https://www.pdpc.gov.sg/help-and-resources/2020/01/model-ai-governance-framework
- https://www.pcpd.org.hk/english/resources_centre/publications/files/ai_protection_framework.pdf
- https://www.globalcompliancenews.com/2023/11/13/https-insightplus-bakermckenzie-com-bm-technology-media-telecommunications_1-thailand-current-state-and-future-direction-of-ai-regulation/
- https://www.tilleke.com/insights/key-concerns-and-provisions-in-thailands-draft-ai-regulation/
- https://www.nortonrosefulbright.com/en/knowledge/publications/f7db14bf/thailand-ai-law-movement
- https://www.nectec.or.th/en/about/news/cabinet-national-ai-strategy.html
