Many technology experts believe that we are on the verge of a significant transformation in which AI will reshape fundamental aspects of our lives and society at large. As AI adoption accelerates and its capabilities expand to encompass increasingly complex tasks that can replace human efforts, a host of novel legal considerations and concerns have emerged.
Consequently, regulatory bodies around the world are confronted with an unprecedented challenge: crafting a governance and regulatory framework for AI products and services. Thailand’s regulatory authorities are also under immense pressure as they undertake this critical mission. Over the past recent years, numerous efforts and working groups have been established to investigate and determine the most effective approach for governing and regulating AI in Thailand.
Notably, the Electronic Transactions Development Agency (ETDA) has recently proposed three key draft laws:
- Draft Artificial Intelligence Innovation Promotion Act (the “Draft Act”)
- Draft Notification of the ETDA regarding AI Sandbox (the “Draft AI Sandbox Notification”)
- Draft Notification of the ETDA regarding AI Risk Assessment (the “Draft AI Risk Assessment Notification”)
Collectively, these drafts are referred to as the “Draft Laws” in this article. The public hearing for these Draft Laws was conducted from 31 March to 14 April 2023.
The Draft Laws introduce several key concepts, including:
- AI Regulatory Sandbox
The Draft Act puts forth the establishment of an AI sandbox under the responsibility of the ETDA, serving as a center for fostering the testing and development of quality and safe AI innovations.
The term “AI Entrepreneur” is defined in the Draft Actas an individual or company involved in the sale ofa product or service related to artificial intelligence, irrespective of whether such person or company receives any form of benefit or compensation or is registered for VAT.
The term “AI Innovation” is defined within the Draft Laws as “software or hardware capable of decision-making or problem-solving that learns from processing data sets through computer algorithms, resulting in either a new finding or an existing finding with significant improvement.”
Those who develop AI products, services, or a product/service incorporating AI Innovation are eligible to apply for participation as well as conduct testing under the sandbox program. At a minimum, the application process will require the applicant to provide information regarding the following aspects of its product or service:
- Human agency and oversight
- Technical robustness and safety
- Privacy and data governance
- Diversity, non-discrimination, and fairness
- Societal and environmental well-being
According to the provisions outlined in the current Draft Act, participation in the sandbox is voluntary. However, participants would have the opportunity to receive support and gain further clarity and insights into the compliance requirements pertaining to their AI product or service. In addition, in specific instances where it is deemed necessary for testing AI, ETDA may engage in discussions with the other relevant Thai regulators to explore the possibility of granting exemptions from certain compliance regulations.
- Data Sharing
Recognizing the crucial role of high-quality training data in AI innovation, the Draft Act stipulates that the ETDA shall actively promote, support, and assist governmental entities, public institutions, and individuals with respect to data sharing for the purpose of innovations. In facilitating the promotion, support, and assistance of the abovementioned parties for the purpose of developing AI innovations, the ETDA will also be authorized to establish manuals, guidelines, or any other means to regulate data sharing. However, these measures must not undermine the free and fair competition among private entities. We anticipate there will be the introduction of subordinate legislations that will provide clearer and more specific details regarding the extent of the ETDA’s regulatory authority concerning data sharing.
Furthermore, the Draft Act includes provisions to govern intermediaries involved in the sharing, exchanging, selling, and buying of data used in AI development. If such an intermediary meets the criteria to be further defined in a sub-regulation by ETDA, then such intermediary would bene subject to registration obligations, along with other requirements such as providing a pledge or insurance coverage for potential harm to data subjects or other affected individuals resulting from the intermediary’s operation.
- AI Standards
The primary objective of the Draft Act is to establish an AI standard certification system, empowering consumers with a reliable tool for assessing the quality and safety of AI products and services available in the market.
Under this system, businesses may submit applications to the ETDA to ascertain whether their AI products or services meet the standards set by ETDA. If deemed eligible, they may be granted a certification mark. It is important to note that obtaining this certification mark is not compulsory for AI providers; rather, it is intended as a voluntary measure to enhance trust among potential customers.
Although the specific AI standards have yet to be defined in the current Draft Laws, the Draft Act proposes that the AI standards encompass the same aspects as proposed for the AI of sandbox applicants, listed in bullet points in 1. AI Regulatory Sandbox above.
- Contract standards for AI products/services
While AI providers are required to comply with the minimum contract standards set by the ETDA, the Draft Act suggests that if there are existing provisions in any other law specifically aimed at consumer protection, the AI provider shall instead comply with those specific provisions in their contracts.
- AI risk assessment and the “AI systems that may be allowed under strict monitoring”
Another objective of the Draft Act is to foster and regulate the safety and reliability of AI products and services. Criteria for conducting risk assessment are defined in the Draft AI Risk Assessment Notification and other forthcoming notifications. Under the Draft Act, the criteria for AI risk assessment must be developed in consideration of potential AI applications that may influence or change human behavior, pose anyone physical or mental harm, or have significant impacts on social or economic stability.
Under the Draft Act, the ETDA has the authority to designate a list of “AI systems that may be allowed under strict monitoring” if deemed necessary. Additionally, the ETDA may consider establishing a list of risk audit and risk management duties for such AI systems. The criteria for conducting risk audit and risk management measures will be further described in a forthcoming notification.
In cases where providers of “AI systems that may be allowed under strict monitoring” operates from outside of Thailand but offer services to users within the country, such providers would be subject to a registration requirement. This entails appointing a local representative and providing them with the necessary documents and information which the AI providers are supposed to disclose under the Draft Act. The appointed local representative must be capable of coordinating with ETDA according to the duties under the Draft Act.
Based on the Draft Laws and the official supplemental documents provided during the public hearing, it is evident that the ETDA is contemplating an approach that fosters innovation through collaboration between the regulator and business owners, utilizing the sandbox framework to jointly explore appropriate regulation. Although the approach seems to be one of promotion rather than regulation, innovators should be aware that ETDA will possess various authorities to issue administrative orders and regulate many different aspects of their operations. if the current versions of the Draft Laws are enacted, there could be many additional compliance requirements to meet.
The Draft Laws are currently in the early stages of the extensive legislative process and are subject to potential revisions. In the meantime, interested persons should stay updated on the general developments as ETDA or other governmental authorities may be working towards regulating and promoting AI innovation in Thailand within the boundaries of their existing authority and applicable laws.
To keep abreast of the latest developments in AI regulation in Thailand, we will be publishing a series of newsletters in the upcoming weeks. Subscribe to our newsletter here, and we will ensure that you receive the most recent updates directly via email as soon as they become available.
Radeemada Mungkarndee, PhD, LLB