July 14, 2026
Thailand’s National Broadcasting and Telecommunications Commission (NBTC) has published guidelines establishing a risk-based framework for the responsible use of artificial intelligence by telecom licensees. Released on July 2, 2026, the Guidelines on the Use of Artificial Intelligence for Telecommunications Services address governance structures, ethical principles, lifecycle management, and consumer protection obligations.
Scope and Legal Context
The nonbinding guidelines apply to holders of telecom business licenses under Thailand’s telecom licensing laws, but only with respect to the use of AI in providing licensed telecom services. Entities without such licenses are not directly subject to the guidelines, though they may be affected as third-party AI solution providers to licensees.
The guidelines supplement and should be read alongside existing laws, including the Cybersecurity Act, the Personal Data Protection Act (PDPA), the Computer Crime Act, and the NBTC Notification regarding Measures to Protect Telecommunications Service Users’ Rights Regarding Personal Data, Privacy Rights, and Freedom of Telecommunications, as well as forthcoming AI governance legislation being drafted by the ETDA.
AI Governance Structure
Licensees are expected to establish committees, working groups, or designated officers at both policy and operational levels to set strategic direction for AI use, formulate governance policies and tools, and oversee risk management. Roles, responsibilities, and accountability should be clearly defined for all personnel across every stage of the AI lifecycle—including for third-party AI solution providers and outsourced service providers, whose obligations should be explicitly documented in service agreements.
Core Principles
The guidelines identify six core principles that licensees should adhere to when deploying AI:
Compliance with laws, ethics, and international standards: AI should respect privacy, dignity, and human rights, and content filtering for inputs and outputs should be considered. For example, the AI should not be designed and developed to be used in generating false information, supporting illegal activities, or causing damage to people.
Fairness: Training data should be diverse,