An Intelligent Approach to AI Governance
The upcoming legislation in Thailand aims to provide protection to users from potential AI risks and remove legal barriers that existing laws cannot address. As the country seeks to establish an AI ecosystem and widen adoption, the Electronic Transactions Development Agency (ETDA) has recently completed an online public hearing on the draft legislation.
Development of Thailand’s AI Regulatory Framework
Sak Segkhoonthod, a senior advisor at ETDA, highlighted that current enforcement of AI rules has been largely based on soft laws or guidelines. The need for a dedicated AI law arises from the necessity to efficiently deal with the impacts of evolving technology.
Since 2022, Thailand has studied global models, particularly the EU’s AI Act, and introduced two draft laws: one focused on regulating AI-enabled business services, and another aimed at promoting AI innovation. These drafts will be combined to form the basis of the new AI law.
Both drafts adopt a risk-based framework, classifying AI systems into prohibited, high risk, and general use categories.
Proposed AI Governance Framework
ETDA proposes a four-tiered approach to promote AI governance:
- First Tier: Collaborate with other countries to enhance Thailand’s global position in AI governance and embrace Unesco’s principles to align with international ethical standards.
- Second Tier: Assign sectoral regulators to oversee policies in their respective domains.
- Third Tier: Encourage corporate implementation through practical tools and guidelines, such as the AI Governance Guidelines for Executives.
- Fourth Tier: Promote AI literacy at the individual level.
Benefits of the AI Law
The proposed legislation seeks to:
- Protect users from potential AI risks.
- Establish governance rules and remove legal barriers that existing laws cannot address, thus facilitating broader AI adoption.
For instance, current regulations from the Transport Ministry do not support the deployment of autonomous vehicles since they were not designed to address uncrewed systems. The new AI law aims to support innovations by efficiently removing regulatory hurdles.
Key Principles in the Draft
The principles focus on supervising AI risks and ensuring that legal recognition is granted to actions and outcomes produced by AI, without denying such recognition solely due to a lack of human intervention. Moreover, all actions derived from AI must remain attributable to humans.
Individuals may be legally exempt from acts generated by AI if the responsible party could not have reasonably foreseen the AI’s behavior, and the other party was aware of the unforeseeable nature of the AI’s actions.
The draft will not define a list of prohibited or high-risk AI applications. Instead, it empowers sectoral regulators to define these lists based on their expertise.
Enforcement of AI Law
The AI Governance Center (AIGC) at ETDA is expected to coordinate with related parties on law enforcement. Existing regulators will define and enforce rules for high-risk AI in their respective sectors. Two key committees will be established under the new law:
- Regulator Committee: Responsible for issuing frameworks and setting policies.
- Expertise Committee: Monitors and evaluates emerging AI risks to ensure timely regulatory responses.
Industry Perspective on the Draft
As of June 20, feedback from 80 organizations, including major players like Google and Microsoft, largely praised the draft for balancing the prohibition of harmful uses with the promotion of innovation. However, concerns were raised regarding the readiness of sectoral regulators to efficiently supervise AI.
Issues of AI sovereignty were also highlighted, noting the risk that foreign generative AI models may provide incomplete or inaccurate responses to users in Thailand due to limited local data representation.
Industry leaders emphasized the importance of defining AI clearly within the legislation to avoid far-reaching implications. They proposed establishing objective criteria for identifying high-risk and prohibited AI applications, promoting a uniform interpretation across the private sector.
Additionally, suggestions included a tiered compliance framework for small and medium-sized enterprises and the development of a formal certification program for AI auditors.
Conclusion
The establishment of an AI incident portal is recommended to create a public repository of AI system failures and rights violations, enabling rapid adaptation and fostering a culture of transparency and trust in AI systems.
As Thailand moves forward with its AI legislation, the balancing act of fostering innovation while ensuring safety and ethical standards will be crucial in shaping the future landscape of AI governance.
