Thailand - AI Innovation Promotion Act

Overview

The Draft Act on the Promotion and Support of Artificial Intelligence Innovation (AI Innovation Promotion Act) is a 2023 Thai draft law that seeks to establish a national legal framework to promote AI research, development and responsible commercial deployment while introducing governance measures to protect individuals and the public interest. The draft aims to balance enabling innovation (including regulatory sandboxes and incentives) with obligations for transparency, risk assessment and data governance administered by the designated competent authority. The instrument originated from national AI policy efforts in 2022–2023 and has been subject to iterative drafting and stakeholder consultation thereafter, with public hearings and policy studies led by the Electronic Transactions Development Agency (ETDA) and other national bodies. For the circulated draft text (Thai, March 29, 2023) see the published copy: Draft Act on the Promotion and Support of Artificial Intelligence Innovation (Thai, March 29, 2023). Further policy material and ETDA publications are available from ETDA's AI Governance Center pages (see Sources table below). (etda.or.th - AIGC)

Definitions

The draft provides technology‑neutral definitions intended to underpin scope and obligations. Core defined terms include "artificial intelligence system," "developer/provider," "user," and risk categories such as "prohibited-risk," "high-risk," and "lower-risk." Definitions are framed to accommodate a range of statistical, machine‑learning and rule‑based automated decision systems and are used to determine applicable compliance obligations, governance requirements and enforcement approaches. The draft's definitional approach is intended to ensure clarity for both public- and private‑sector actors. (See ETDA materials for related definitional discussion: etda.or.th - AIGC)

Governance and Institutional Framework

The draft envisages a competent national authority (in practice ETDA and its AI Governance Center have been central to policy development) empowered to supervise implementation, publish technical standards and guidance, operate sandboxes, coordinate with sectoral regulators, and issue subordinate regulations and lists (for example, lists of high‑risk uses). Advisory and multi‑stakeholder arrangements are anticipated to align the instrument with national AI strategy and international norms. The competent authority is expected to administer registration, conformity assessment, enforcement, and the promotion/support measures envisioned by the draft. (etda.or.th - AIGC)

Key Focus Areas

• Risk-based classification: The draft adopts a risk‑based approach that distinguishes between prohibited‑risk, high‑risk and lower‑risk AI systems to focus regulatory measures where societal, economic or security impacts are greatest. (etda.or.th - AIGC)
• Transparency and documentation: Requirements for transparency, technical documentation, logging and model cards or technical files are envisaged, particularly for higher‑risk systems.
• Impact assessment and mitigation: Pre‑deployment impact assessments, risk assessments and mitigation measures are proposed for high‑risk AI deployments.
• Registration, conformity and notifications: The draft foresees registration, conformity assessment or notification regimes for specified AI services and functions, as determined by delegated lists published by the competent authority.
• Innovation facilitation: Encouragement measures include establishment and operation of AI testing centres or regulatory sandboxes, R&D incentives, and limited legal safe‑harbors for activities such as text‑and‑data mining to support innovation and experimentation.
• Delegation and subordinate regulation: The competent authority is granted powers to issue subsidiary rules, technical standards and lists (for example, lists of standardized high‑risk uses), enabling iterative regulatory detail to be set by administrative instrument.

Implementation Framework

Implementation is to be coordinated by the competent authority through subordinate regulations, technical standards and guidance. Compliance elements described in the draft include pre‑deployment impact assessments for high‑risk systems, record‑keeping and documentation (logs, model cards, technical files), registration or notification for specified categories, adherence to privacy safeguards under Thailand's Personal Data Protection Act (PDPA) when personal data are processed, and participation options for experimental deployments in regulated sandboxes. Standards alignment and conformity assessment mechanisms are foreseen to support market trust and facilitate supervised market entry for novel AI services. Sectoral regulators and existing law (notably the PDPA) remain relevant and complementary to this framework. (etda.or.th - Regulator / Public Hearing)

Monitoring and Evaluation

The draft provides supervisory powers for the competent authority to monitor compliance, investigate potential breaches, require corrective measures, suspend or restrict unlawful AI operations, and impose administrative sanctions or penalties where appropriate. Enforcement is designed to be proportionate, with expectations that subordinate legislation will detail specific penalties and procedural safeguards. Monitoring and evaluation activities are likely to include the review of sandbox outcomes, conformity assessment results, and coordinated oversight with sectoral regulators to assess the impact of the regime on innovation and public interest outcomes. (For examples of ETDA's governance activities and public consultation materials see Sources.)

Penalties, Liability, and Appeals

The draft contemplates administrative enforcement tools including corrective orders, temporary suspension or restriction of operations, and administrative sanctions for non‑compliance. Specific penalties, liability rules and appeal procedures are to be detailed in subordinate legislation and procedural provisions; the draft emphasises proportional enforcement to protect public interest while avoiding undue barriers to innovation. Procedural safeguards and appeals mechanisms are expected to be established in implementing regulations issued by the competent authority. (etda.or.th - Regulator / Public Hearing)

Relationship to Other Instruments

The draft is intended to operate alongside and complement existing Thai law, notably the Personal Data Protection Act (PDPA), sectoral regulatory regimes, national AI strategic plans and digital platform rules (including related draft decrees on platform and AI business activities). It envisages active coordination with sectoral regulators to avoid duplication and to ensure coherent oversight. The draft also signals interaction with other national instruments and policy processes that have been part of public consultations and national consolidation efforts. (library.parliament.go.th)

International Alignment

While the draft is focused on establishing a domestic framework for Thai AI innovation and governance, it anticipates coordination with international norms and standards through advisory and multi‑stakeholder arrangements and by publishing technical standards and guidance that take account of global best practices. The competent authority's role in aligning standards and conformity assessment mechanisms is intended to support interoperability and international cooperation. (See ETDA publications and national strategy materials for further context.)

Implementation Timeline

Sources and References