The European Union’s pioneering approach to regulating Artificial Intelligence (AI) has set a new global standard, with implications reaching far beyond its borders. On 21 May 2024, the European Council formally adopted the EU AI Act, scheduled for full enforcement by 2 August 2026. This landmark legislation aims to mitigate potential harm from AI usage while fostering innovation.

The EU AI Act employs a risk-based approach, categorizing AI systems into four levels:

1. Unacceptable Risk: Prohibited AI systems that pose threats to human rights, such as social scoring systems and real-time face recognition systems.
2. High-Risk: AI systems are required to meet specific requirements and undergo conformity assessments, including biometric identification and critical infrastructure systems.
3. Limited-Risk: AI systems that must fulfill obligations before market entry, such as deepfakes and chatbots.
4. Minimal-Risk: AI systems that must adhere to a code of conduct, including speech recognition and spam filtering systems.

The EU AI Act also introduces AI Regulatory Sandboxes, monitored by National Competent Authorities, to ensure compliance before market deployment.

Impact on Thailand

While Thailand is not an EU member, the EU AI Act’s influence extends globally, presenting both challenges and opportunities:

1. Legal Framework Development: The Act provides a blueprint for Thailand to develop robust AI laws and regulations.
2. Business Performance Enhancement: Thai AI-related businesses may need to elevate their standards to operate in the EU market, indirectly improving the domestic AI industry.
3. Access to Advanced AI Systems: Thailand may benefit from the influx of high-standard AI systems developed under EU regulations.

Thai AI Legal Instruments in Development

Influenced by the EU AI Act, Thailand is currently drafting three main legal instruments:

1. The Draft Act on the Promotion and Support of AI Innovations in Thailand: This act establishes general rules, requirements, and authorities for AI control. While influenced by the EU AI Act, it relies more on subordinate laws for specific requirements. Uniquely, it includes provisions for reimbursing damage caused by AI use when no responsible party can be identified.
2. The Draft Royal Decree on Business Operations that Use Artificial Intelligent Systems: This decree adopts the EU’s risk-based approach and requirements for High-Risk AI. However, it defers detailed explanations to sub-regulations. It also includes administrative and criminal penalties for non-compliance.
3. The Draft Notification of Electronic Transactions Development Agency Re: AI Sandbox: This notification focuses on AI Sandboxes, crucial for pre-market implementation testing. Unlike the EU’s mandatory approach, Thailand’s AI Sandbox is voluntary.

Key Differences and Adaptations

While heavily influenced by the EU AI Act, Thailand’s approach shows some notable differences:

1. Regulatory Depth: Thai drafts often defer detailed requirements to subordinate laws, whereas the EU AI Act provides comprehensive explanations within the main legislation.
2. Enforcement Approach: Thailand includes specific provisions for penalties and damage reimbursement, which are not as explicitly outlined in the EU AI Act.
3. Sandbox Implementation: Thailand opts for a voluntary AI Sandbox approach, contrasting with the EU’s mandatory system.

Conclusion

The EU AI Act marks a significant milestone in AI regulation, influencing global approaches including Thailand. While Thailand is not obligated to follow EU standards, the similarities in their developing legal instruments highlight the EU AI Act’s far-reaching impact. As Thailand continues to refine its AI legal framework, it balances adopting international best practices with tailoring regulations to its specific needs and context.

As the global AI landscape evolves, Thailand’s proactive approach in developing comprehensive AI regulations is to navigate the challenges and opportunities presented by this transformative technology.

Key Takeaways

• EU has passed the first AI Act which will be fully enforced by 2 August 2026.
• The EU AI Act uses risk-based approaches to prevent the possible harm to human.
• The EU AI Act does affect Thailand in the field of commercialization and legalization.
• Thailand does not have the burden to follow EU AI Act. However, Thailand heavily influenced by EU AI Act in the process of drafting.