Circular 05/2026/TT-BKHCN: National Artificial Intelligence Ethics Framework
RegHub explainer by New MarketerLast updated:
Based on:Thông tư số 05/2026/TT-BKHCN ngày 10 tháng 3 năm 2026 - Government Official Gazette
This explanation was generated by AI and checked by an automated AI review, not by a human expert. It is not legal, tax or accounting advice and may contain errors. Check the official document before you rely on it.
The Ministry of Science and Technology issued Circular No. 05/2026/TT-BKHCN dated 10 March 2026, promulgating the National Artificial Intelligence Ethics Framework. It applies to state management agencies and organisations or individuals acting as developers, providers, deployers, or users of AI systems in the context of state administration or public services. The Framework sets out three principle clusters: (1) Safety, reliability, and harm avoidance — requiring safety-by-design, human oversight, and system security; (2) Respect for human and civil rights, fairness, transparency, and non-discrimination; (3) Promotion of well-being, prosperity, and sustainable development. Non-state organisations are encouraged (but not required) to comply. This document does not directly create tax or financial obligations for SMEs and is therefore of low relevance to an SME tax and regulatory intelligence platform.
Circular 05/2026/TT-BKHCN — National AI Ethics Framework
Issuing authority: Ministry of Science and Technology
Date issued: 10 March 2026
Primary legal basis: Artificial Intelligence Law No. 134/2025/QH15
Applicability
- Mandatory: State management agencies and organisations/individuals who develop, provide, deploy, or use AI systems for state administration or public services
- Encouraged: Vietnamese and foreign organisations and individuals conducting AI research, development, or use in Vietnam
Three Clusters of AI Ethics Principles
1. Safety, Reliability, and Harm Avoidance
- Safety by design: identify harmful scenarios in advance, build preventive measures
- Reliability and quality: quality criteria for data, models, and outputs; internal testing before deployment
- Human oversight: supervision and intervention mechanisms proportionate to system impact
- Resilience and security: incident-reporting mechanisms, contingency plans
- AI system security: protection against data poisoning, model poisoning, adversarial attacks
2. Human Rights, Fairness, Transparency, and Non-Discrimination
- Respect privacy, personal data, freedom of will, and right to information
- Fairness: identify and mitigate data bias, model bias, and operational bias
- Transparency: notify users of AI use; provide reasonable information about the system
- Explainability and accountability: clearly assign responsibility for AI-driven decisions
3. Well-being, Prosperity, and Sustainable Development
- Social benefit: identify public benefits, address negative impacts before deployment
- Inclusive development: accessible interfaces, bridging the digital divide
- Sustainable development: energy-efficient solutions, reduced emissions across the system lifecycle
- Respect Vietnamese cultural values and social norms