With ambitions to become not just AI adopters but leaders in research and development, countries ranging from technologically advanced economies such as Japan, South Korea, China and Singapore to populous emerging markets such as India and Indonesia are refining their regulatory frameworks.
From developed nations like Australia and New Zealand to middle-income economies such as Thailand and Vietnam, governments are seeking a delicate balance that encourages innovation while preventing the worst of AI harms.
How they achieve this can be as varied as their legal systems and levels of economic development. Given the stark differences, each country’s definition of balance between innovation and protection will land at a different point on the regulatory spectrum.
What is clear, though, is that Asia-Pacific countries are no longer just passive tech consumers or mere manufacturing hubs, and their governments are no longer just policy takers.
With some $90 billion of investment into AI and data centers in the region announced by tech giants in recent years — led by Southeast Asia, Japan and India — the Asia-Pacific is a global force in the future of AI.
— Defined directions —
It is perhaps not surprising to see that the region’s more technologically advanced countries are the ones with more defined approaches to AI regulation.
Among them, Japan has so far taken one of the hardest stances. In June last year, its Personal Information Protection Commission became the second data-protection regulator in the world to take action against OpenAI, instructing it not to collect sensitive user information without consent for training its ChatGPT tool's machine learning.
This year, it shifted from a soft-law approach to pursuing hard legislation due to rising awareness of AI-related risks, such as the increasing use of AI-generated celebrity images in fraudulent schemes.
In February, ruling-party lawmakers proposed legislation aimed at maximizing AI benefits while minimizing risks, particularly targeting developers of large AI foundation models. By August, the government had established a panel to explore potential AI legislation.
Similarly, South Korea is moving toward a comprehensive AI law while also pursuing industry-specific regulations. Multiple draft bills adopt principles-based approaches, including one from the government. A presidential AI committee has also been created to accelerate policy decision making, with a primary focus on promoting industry development.
South Korea boasts a comprehensive AI ecosystem, including a tech-forward government, high-end hardware manufacturers, large-scale model developers and a tech-savvy population. The country has officially set its sights on becoming an AI leader, second only to the US and China.
At the other end of the spectrum is Singapore, which leads the Asia-Pacific region in AI readiness but has no plans for a comprehensive law. Instead, it follows a light-touch approach with guidelines and targeted regulations, such as a recent law banning deepfakes in elections.
At the same time, it is collaborating closely with industry players to thoroughly identify and understand the risks posed by AI before developing and validating appropriate risk-mitigating measures. This business-friendly approach has been hailed as a model that various industry players say they hope other countries will follow.
Table of Asia-Pacific investments
In other parts of the region, regulatory approaches are evolving as quickly as the technology itself. Countries that initially aimed for comprehensive AI laws have scaled back their ambitions, largely due to pushback from industry stakeholders — an indication of the weight that governments place on encouraging AI innovation and adoption alongside protecting against its harms.
China initially included AI legislation in its annual work plan, but progress slowed after industry opposition to two academic proposals that were meant to guide regulators. Industry leaders argue that China’s top priority should be AI development at this stage, especially given its restricted access to advanced US technologies.
Amid this tech rivalry with the US, China’s top legislature has shifted to a more cautious and gradual approach, advocating for consensus-building in AI regulation. For the Chinese government, AI is a “new productive force” (a catchphrase of the government’s latest economic policy) that can boost the competitiveness of industries, and its potential is being harnessed with careful consideration for its impact on employment.
Thailand, Southeast Asia’s second-largest economy, has similarly scaled back its regulatory ambitions. Initially, it seemed poised to adopt a comprehensive AI law modelled on the EU's AI Act. However, recent months have seen regulators call for more studies, seeking an approach that they say would be fair to both businesses and the public.
Conversely, India, which leads the region in AI startups, has reversed its initial opposition to regulating AI to avoid mistakes it made in regulating the Internet. The government also said it wants to hold social media platforms accountable for safety and trust.
The government has said, however, that it will not introduce a standalone law, instead planning to regulate AI through the forthcoming Digital India Act, replacing the Information Technology Act of 2000.
Vietnam, like India, is incorporating AI regulations into its draft Digital Technology Law. Early drafts followed the EU’s risk-based approach, prohibiting AI activities such as behavior manipulation and discrimination. But the latest draft does away with this and instead says an AI system would be considered high-risk only if it meets a narrow set of criteria.
— Still studying —
Other countries are still trying to figure out what the most appropriate approach would be for their circumstances.
Australia, for example, has proposed 10 mandatory guardrails for AI. The government plans to consult on these before deciding whether to adopt AI-specific legislation or update current laws.
As in other countries, the government’s plans to regulate AI have faced industry pushback, with lobbyists warning that over-regulation can stifle the productivity and economic opportunity AI presents.
New Zealand has also published a paper seeking agreement on a strategic approach to AI, but it has so far said it is not considering standalone AI laws.
Indonesia, meanwhile, said it is drafting a ministerial regulation on AI that could be the basis of legislation later. There has been little indication of what approach it would take, but the country’s information technology minister said last April that they are looking at a combination of European and American regulatory approaches to find the most suitable option for the country.
Malaysia just recently launched its national guidelines on artificial intelligence governance and ethics and has not expressed any intentions so far to come out with a comprehensive law. But it pointed out that it has several laws relevant to the adoption of AI principles.
— All about risk —
Even though countries are not rushing to follow the EU's AI Act in the same way that the General Data Protection Regulation, or GDPR, was imported by several Asian governments, many are looking to its risk-based approach to regulating AI.
In August, for example, Japan's then-prime minister Fumio Kishida asked a panel of experts set up to discuss potential legislation on AI to pursue a risk-based approach to addressing AI risks.
Australia similarly plans to adopt a risk-based approach to its planned mandatory guardrails for the use of AI in high-risk settings and, in designing a risk-based regulatory regime for AI, “government will consider the balance of ex-ante and ex-post regulatory measures to effectively target and mitigate known risks.”
In New Zealand, a cabinet paper published in July said the government will pursue a “light-touch, risk-based approach to AI regulation.” India also proposes to regulate AI through the lens of a "risk-based framework," thereby promoting innovation.
The resulting frameworks or regulations would likely be different, however, because each country’s definition of low risk, high risk and unacceptable risk would vary based on its specific motivations and priorities.
Ultimately, the AI regulatory landscape in the Asia-Pacific is shaping up to be as diverse as the region itself.
— With reporting by Yonnex Li, Freny Patel, Sachiko Sakamaki, Toko Sekiguchi, Saloni Sinha, Olivia Wang and Choonsik Yoo.
Please email editors@mlex.com to contact the editorial staff regarding this story, or to submit the names of lawyers and advisers.