Thailand Moves Forward with Draft Artificial Intelligence Act in 2026 On 9


On 9 July 2026, Electronic Transactions Development Agency (ETDA) published the Draft Artificial Intelligence (AI) Act for public consultation through the Law Portal of Office of the Council of State (OCS). The public hearing has been closed on 14 August 2026, allowing stakeholders and general public to submit comments and feedback on the proposed legislation.
The Draft AI Act categorizes the risks associated with the use of AI and sets out the duties and obligations of service providers, deployers, and providers located outside Thailand, as detailed in the table below.

1. AI Risk Classification under The Current Draft AI Act (Risk-Based Framework)
The draft Act sorts AI systems into 2 tiers of regulatory attention i.e., Prohibited (Unacceptable) Risk AI and High Risk AI, each with distinct obligations and oversight mechanisms.

Business takeaway: Failure to implement adequate risk management measures for High-Risk AI can result in an administrative fine of up to THB 1 million.
Remark: This is only a preliminary draft of the legislation and may be subject to further amendments in the future.

2. Serious AI Incident Duties for Businesses - Direct Obligations for Providers & Deployers
  • Actual or Potential Serious AI Incidents
The Authority may order:
(1) Temporary suspension of AI services or product recalls by providers; or
(2) Suspension of AI deployment by deployers

Where an actual or potential serious incident threatens national security, economic stability, or public safety.
 
  • Potential Serious AI Incidents
Providers and deployers must jointly:
(1) Assess the risk and potential impact of the serious AI incident;
(2) Identify and implement risk mitigation measures; and
(3) Promptly notify the relevant sectoral regulator and the Office.

Applies to all AI systems, including prohibited-risk, high-risk, and other AI systems.
 
  • Escalation for Non-Compliance
  • Non-compliance with orders → court-ordered for suspend of services or recall the products (temporary)
  • AI system on a platform: continued breach → court order to block the dissemination of AI system to deployers within Thailand

Business takeaway: Build incident-detection and joint reporting protocols with your AI supply-chain partners now — breach of these duties carries an administrative fine of up to THB 5 million (approximately USD 150,000).
Furthermore, AI systems that can generate or modify content that may appear to be genuine to the public, such as deepfakes and chatbots, must also comply with the applicable legal requirements set out below.



Transparency Obligations for AI Systems: Deepfakes and Chatbots
  • Developers of AI systems capable of generating or modifying images, audio, videos, or other content that could mislead the public into believing such content depicts real persons, events, or facts (e.g., deepfakes) must:
  • Conduct a risk assessment
  • Implement risk mitigation measures
  • Adopt technical measures to ensure AI-generated or AI-modified content is identifiable, including, at a minimum, the use of a machine-readable mark
  • Persons who publish AI-generated or AI-modified images, audio, or videos that could mislead the public on specified matters must clearly disclose that the content was generated or modified by AI.
  • Developers, deployers, and platform service providers must prepare and retain records summarising their perfomrance and update their AI-generated content risk management measures at least annually.

Legal sanction
  • 1. Civil liability
The draft AI Act establishes a strict liability regime, under which persons are jointly liable for damages arising from violations of the AI Act, irrespective of intent or negligence.
  • 2. Administrative fine
The applicable administrative fine ranges from THB 1 million to THB 5 million.
 
Download