In short
The Ministry of Science and Expertise (MST) has simply launched a draft of the Regulation on Synthetic Intelligence (“AI Regulation“) for public session till 20 October 2025.
Key takeaways
Beneath is our abstract of key points below the draft AI Regulation:
- Implementation timeline: The draft AI Regulation is scheduled to enter into pressure on 1 January 2026, superseding AI guidelines within the Digital Expertise Trade Regulation. Nevertheless, the regulation will likely be rolled out in phases: the provisions on prohibited acts and regulatory sandbox will apply after 12 months from the regulation’s efficient date, whereas the obligations for high-risk AI techniques will solely take impact after 18 months.
- Foundational ideas: All actions associated to AI techniques in Vietnam should adhere to ideas akin to equity and non-discrimination, transparency and explainability, human centrism, safety and security, and accountability.
- Danger-based classification and governance: AI techniques are categorized into 4 ranges: unacceptable (prohibited), excessive (strict obligations like assessments and registration), medium (transparency necessities), and low (voluntary requirements) dangers. Companies in sectors like finance, well being, or justice face heightened scrutiny.
- Normal-purpose AI (GPAI) mannequin: Modeled after EU ideas, GPAI suppliers should meet primary obligations (e.g., documentation, IP compliance) and enhanced ones for systemic dangers (e.g., evaluations, cybersecurity). This may have impacts on tech companies integrating massive fashions.
- Function-driven accountability: Provide chain stakeholders, together with AI builders, suppliers, deployers, and customers, could maintain completely different obligations relying on their precise function and management over an AI system, which will likely be additional elaborated by the Authorities below a subordinate authorized instrument. Joint legal responsibility for compensation may come up the place the roles of the events can’t be clearly delineated.
- Incentives and Helps for Innovation: Establishes a Nationwide AI Improvement Fund, sandboxes for startups/SMEs, AI clusters, and tax incentives – providing alternatives for home and international companies to spend money on R&D and infrastructure.
- Income-based penalty and different enforcement measures: In circumstances of great violations dedicated by a large-scale group, the financial fantastic imposed may very well be calculated based mostly on a proportion of the group’s world income from the earlier fiscal yr. Measures akin to non permanent termination or everlasting prohibition of an AI system will also be utilized if the system is discovered to trigger a severe nationwide safety danger, amongst different situations.
Being quick tracked to fulfill this yr’s deadline, the draft AI Regulation will likely be mentioned and handed by the Nationwide Meeting in its upcoming assembly session ranging from 20 October 2025. Time is of the essence right here – stakeholders have to intently observe the drafting progress and shortly submit their feedback to the lawmakers.
* * * * *
Alex Do, IPTech Government and Danny Truong, Trainee have contributed to this authorized replace.
The put up Vietnam: Releasing draft AI Regulation for complete AI governance framework appeared first on International Compliance Information.
