# Vietnam to Update AI Strategy and Enact AI Law by 2025, Building National AI Infrastructure

> Vietnam to update its AI strategy and pass an AI Law by end-2025, with AI infrastructure, shared data, and human oversight. Counsel: prep for tighter rules on data, risk, security.

- Type: AI news
- Page: https://completeaitraining.com/news/vietnam-to-update-ai-strategy-and-enact-ai-law-by-2025/
- Published: 2025-09-21
- Updated: 2025-09-21
- Author: Jeroen Erne
- Jobs: Legal
- Reading time: 3 min

### Vietnam to update national AI strategy and AI Law by late 2025: what legal teams need to know

Vietnam plans to publish an updated national AI strategy and a new AI Law by the end of 2025. The policy line is clear: treat AI as core infrastructure, build national compute and data assets, and raise competitiveness while keeping humans in control.

Senior officials have flagged a national AI supercomputing center, shared open AI data, and a risk-aware approach that keeps decisions with people. Advisors also point to "technology diplomacy" and "data diplomacy," signaling cross-border cooperation and standards will matter as much as local rules.

#### Why this matters for counsel

The 2021 AI strategy is being refreshed to match generative AI at scale and intensifying international competition. Expect a broader legal perimeter: data governance, safety, accountability, cybersecurity, procurement, and sector-specific rules tied together by a national data architecture.

Vietnam is already among ASEAN's leaders in AI readiness and adoption. A formal AI Law will set expectations for how organizations build, buy, deploy, and audit AI systems in both public and private sectors.

#### Signals from officials: emerging pillars of the AI framework

- **AI as infrastructure:** National supercomputing capacity and common data resources to lower entry barriers and standardize practices.
- **Data development and architecture:** Clear guidance for ministries and agencies on system design, connectivity, metadata, and access to shared datasets.
- **Human oversight:** AI as an assistant; humans remain the decision-makers with defined accountability.
- **Alliances and interoperability:** Technology and data diplomacy to enable cooperation, standards alignment, and lawful data sharing.
- **Fast, safe, humane deployment:** Risk awareness, security-by-design, testing, and monitoring likely to be front and center.

#### Rising risk: AI-driven cyberattacks

At a recent summit in Ho Chi Minh City, hundreds of security leaders exchanged methods and tools for AI in information security-on both the defense and attack sides. This trend points to stricter expectations around red-teaming, incident response, logging, and supplier assurance for AI systems.

#### What to prepare now

- **Map your AI footprint:** Inventory models, use cases, datasets, third-party providers, and data flows (training, fine-tuning, inference).
- **Classify risks:** Flag high-impact uses (e.g., employment, credit, healthcare, public services). Require pre-deployment assessments and approvals.
- **Data governance:** Define legal bases for collection and use, consents where required, minimization, retention, and access controls. Track synthetic data usage and provenance.
- **Model lifecycle controls:** Establish testing, bias and performance metrics, explainability standards where feasible, monitoring, and rollback plans.
- **Security-by-design:** Threat models for prompt injection, data leakage, model theft, supply-chain risks, and shadow AI. Mandate logging and incident reporting.
- **Human oversight and accountability:** Specify decision rights, human review thresholds, and escalation. Document who is responsible for outcomes.
- **Contract terms with AI vendors:** Training data rights and restrictions, IP ownership of outputs, confidentiality, audit rights, security controls, bias testing, uptime/SLA, incident notice, and indemnities (IP infringement and security).
- **Procurement and audits:** Prepare for public-sector style due diligence: data sheets, model cards, evaluation reports, and conformity evidence.
- **Cross-border data:** Plan for transfer impact assessments, localization requirements (if any emerge), and processor/sub-processor transparency.
- **Policy and training:** Clear acceptable use policies, approval gates for new tools, and role-based training for developers, product owners, and reviewers.
- **Engage early:** Participate in consultations, pilot sandboxes, and standards groups. Align internal policies with recognized frameworks like the [OECD AI Principles](https://oecd.ai/en/ai-principles?ref=completeaitraining.com).

#### Sector angles to watch

- **Financial services:** Model risk management, explainability, fairness, and consumer disclosure will draw scrutiny.
- **Healthcare:** Data consent, clinical validation, and post-market surveillance for AI-enabled tools.
- **Employment and education:** Bias controls, transparency to affected individuals, and appeal mechanisms for automated decisions.
- **Public sector:** Procurement conditions, logging, auditability, and responsible use of national datasets.

The message from leadership is consistent: scale AI access, build shared infrastructure, and keep people in charge. For legal teams, that translates into clear accountability frameworks, audit-ready documentation, and disciplined vendor management.

#### Timeline and next steps

The updated strategy and AI Law are slated for release by end-2025. Use the lead time to close gaps: inventory, risk classification, contracts, security controls, and governance committees. Early movers will have fewer surprises when the law lands.

If you need a fast way to brief non-legal stakeholders on AI capabilities by role, see this curated index: [AI courses by job](https://completeaitraining.com/courses-by-job/).

---

This is the Markdown form of https://completeaitraining.com/news/vietnam-to-update-ai-strategy-and-enact-ai-law-by-2025/. The whole site is mapped at https://completeaitraining.com/llms.txt.
