Thailand - AI Ethics Guidelines

Overview

The Digital Thailand AI Ethics Guideline is a national, non‑binding ethics framework published to guide the design, development, deployment and oversight of AI systems across Thailand’s public and private sectors. It sets six core principles (competitiveness & sustainability; law, ethics & international standards; transparency & accountability; security & privacy; fairness & inclusiveness; reliability) and provides practical recommendations for regulators, developers, providers and users to translate those principles into operational practice. The document situates the principles within Thailand’s broader National AI Strategy and cross‑references data protection requirements under the Personal Data Protection Act (PDPA). The full guideline (Thai language PDF) is available from the Electronic Transactions Development Agency (ETDA) and is the primary source for the framework. Digital Thailand - AI Ethics Guideline (ETDA PDF). ([etda.or.th](https://www.etda.or.th/getattachment/9d370f25-f37a-4b7c-b661-48d2d730651d/Digital-Thailand-AI-Ethics-Principle-and-Guideline.pdf.aspx?lang=th-TH))

Definitions

The Guidelines provide an operational glossary that defines key terms such as "AI", "researcher", "designer", "developer", "provider", "user", "transparency", "accountability", "traceability", "security", "privacy", "fairness" and "reliability" to ensure a common understanding among stakeholders. Definitions emphasize lifecycle perspectives (research → design → deployment → monitoring) and distinguish roles and responsibilities for actors across public and private sectors. This shared terminology is intended to reduce ambiguity when applying the principles in procurement, regulation and service delivery. ([etda.or.th](https://www.etda.or.th/getattachment/9d370f25-f37a-4b7c-b661-48d2d730651d/Digital-Thailand-AI-Ethics-Principle-and-Guideline.pdf.aspx?lang=th-TH))

Governance and Institutional Framework

The Guidelines recommend a multi‑layered governance architecture that includes: (1) national coordination (e.g., a National AI Committee aligned with the Cabinet and supported by lead ministries such as the Ministry of Digital Economy and Society and MHESI); (2) sectoral supervisory authorities that adapt the principles to sector risks; and (3) organizational governance inside providers (AI ethics committees, risk units, and independent advisory panels). They call for clear allocation of responsibilities for oversight, registration or notification of high‑risk systems (where applicable), and the promotion of open platforms and public testing facilities. The Guidelines also urge capacity building for civil servants and the creation of public registries or disclosure mechanisms to support accountability. For background and national alignment see the Thailand National AI Strategy and Action Plan (2022–2027). Thailand National AI Strategy & Action Plan. ([ai.in.th](https://www.ai.in.th/en/about-ai-thailand/?utm_source=openai))

Key Focus Areas

The Guidelines translate high‑level principles into six focus areas: (1) legal & ethical compliance — ensuring AI respects laws and human rights; (2) transparency & accountability — recommending documentation, model cards and decision review channels; (3) data governance & privacy — aligning AI data use with PDPA obligations and privacy‑by‑design; (4) system safety & reliability — advocating testing, robustness evaluation and continuous monitoring; (5) fairness & inclusion — requiring bias audits, representative data and mitigation strategies; and (6) security & resilience — specifying cybersecurity practices for models and datasets. For each area, the Guidelines propose concrete practices such as audit trails, performance measurement, exception handling, channels for user feedback and processes for human intervention in automated decisions. These focus areas are repeatedly emphasized as essential to trustworthiness and to maintaining competitive national AI capability while minimizing social harms. The Guidelines also highlight the need for explainability proportionate to impact and for accessible communications to users about system capabilities and limitations. ([etda.or.th](https://www.etda.or.th/getattachment/9d370f25-f37a-4b7c-b661-48d2d730651d/Digital-Thailand-AI-Ethics-Principle-and-Guideline.pdf.aspx?lang=th-TH))

Implementation Framework

Implementation guidance is practical and phased: (a) adopt governance structures (ethics boards, appoint accountable officers); (b) establish processes for AI impact assessments and risk categorization; (c) require testing and validation protocols for accuracy, repeatability and robustness; (d) integrate privacy/security by design; (e) maintain records and documentation (audit logs, datasets provenance, model versions); and (f) build user‑facing redress and feedback channels. The Guidelines recommend that procurement and public sector deployments adopt the same standards to drive market incentives toward trustworthy AI. They also advise regulators to coordinate internationally when developing binding measures. ([etda.or.th](https://www.etda.or.th/getattachment/9d370f25-f37a-4b7c-b661-48d2d730651d/Digital-Thailand-AI-Ethics-Principle-and-Guideline.pdf.aspx?lang=th-TH))

Monitoring and Evaluation

Monitoring approaches include continuous post‑deployment surveillance, periodic audits, performance benchmarking and user reporting mechanisms. The Guidelines encourage both internal governance monitoring and external review (third‑party audits, independent oversight panels). Indicator examples include error rates, bias metrics, uptime/availability, data provenance completeness and user complaint resolution times. They recommend public reporting cycles and mechanisms to update practices in light of evidence and technological change. The document stresses iterative evaluation rather than one‑time certification for systems in dynamic operational contexts. ([etda.or.th](https://www.etda.or.th/getattachment/9d370f25-f37a-4b7c-b661-48d2d730651d/Digital-Thailand-AI-Ethics-Principle-and-Guideline.pdf.aspx?lang=th-TH))

Penalties, Liability, and Appeals

As a soft‑law instrument, the Guidelines themselves do not create new criminal penalties; however, they expressly link compliance expectations to existing Thai laws (for example PDPA obligations and sectoral safety or consumer protection laws). Where violations implicate statutory protections (data breaches, discrimination, safety harms), statutory remedies and penalties (including administrative fines, civil liability and criminal sanctions where provided by law) will apply. The Guidelines recommend clear internal redress, escalation procedures and access to external appeals or supervisory complaints. Organizations are advised to align their contractual and insurance arrangements to reflect potential legal liabilities for AI‑caused harms. ([etda.or.th](https://www.etda.or.th/getattachment/9d370f25-f37a-4b7c-b661-48d2d730651d/Digital-Thailand-AI-Ethics-Principle-and-Guideline.pdf.aspx?lang=th-TH))

Relationship to Other Instruments

The Guidelines are positioned as a national ethics framework intended to inform and be interoperable with (a) the Thailand National AI Strategy & Action Plan (2022–2027), (b) the PDPA and other Thai legislation, (c) sectoral regulatory instruments, and (d) international instruments such as OECD and UNESCO AI principles. They are therefore designed to be referenced by subsequent binding regulations (for example, draft royal decrees and sector rules) and by standards bodies as a common ethical baseline. The Guidelines also recommend that regulators consider conformity assessment, voluntary registration and certification schemes for high‑risk systems. ([ai.in.th](https://www.ai.in.th/en/about-ai-thailand/?utm_source=openai))

International Alignment

The Guidelines draw explicitly on international instruments (OECD, EU and UNESCO guidance) to promote cross‑border interoperability and to support trade and research collaboration. They encourage adherence to international standards (technical and normative) and recommend participation in global cooperative mechanisms for AI testing, benchmarking and standardization. The document frames Thailand’s approach as one that aims to align national policy with global best practice while retaining policy space for country‑specific measures. OECD AI Policy Observatory provides a complementary international reference. ([etda.or.th](https://www.etda.or.th/getattachment/9d370f25-f37a-4b7c-b661-48d2d730651d/Digital-Thailand-AI-Ethics-Principle-and-Guideline.pdf.aspx?lang=th-TH))

Implementation Timeline

Milestone	Date	Notes
Guidelines published by ETDA	2022-03-22	National publication / soft‑law dissemination. ETDA PDF
Cabinet approval of National AI Strategy	2022-07-26	National AI Strategy & Action Plan (2022–2027) adoption. AI Thailand
PDPA initial enforcement (context)	2019-05-24	Personal Data Protection Act effective date (public law referenced in guideline). PDPA (Royal Gazette)

Compliance Checklist

Item	Action
Principles adopted	Record organizational policy referencing the six AI ethics principles.
Governance	Establish an AI ethics committee and appoint accountable officer(s).
Risk assessment	Conduct AI impact/risk assessments before deployment.
Data protection	Ensure PDPA compliance; document lawful bases and consents.
Transparency	Publish model cards, purpose statements, and user notices proportionate to impact.
Testing & monitoring	Implement pre‑deployment testing, continuous monitoring and incident response processes.
Audit & records	Maintain audit trails, dataset provenance and version control.
User redress	Provide accessible feedback and decision‑review channels.

Sources and References

Source	Type
Digital Thailand - AI Ethics Guideline (ETDA PDF)	Primary Source
Digital Policy Alert — Published AI Ethics Guidelines for Digital Thailand	Secondary Source
Thailand National AI Strategy & Action Plan (AI Thailand)	Primary/National Strategy
Personal Data Protection Act B.E. 2562 (2019) — Royal Gazette	Primary Source