Vietnam Proposes ASEAN–China AI Pact on Governance Safety and Skills
At the 2026 China–ASEAN Ministerial Roundtable on Artificial Intelligence in Nanning on September 17, 2026, Vietnamese Deputy Minister of Science and Technology Hoàng Minh proposed three practical cooperation tracks with China and ASEAN peers: deeper policy dialogue on governance approaches, joint AI safety research testing and assessment, and expanded human-resource development to narrow capability gaps. He told delegates a safe AI future must rest on each country’s governance capacity and mutual trust.
Filed under Policy and dated September 18, 2026, this AI4Vietnam briefing treats the speech as Vietnamese diplomacy-and-standards news rather than a bilateral tech procurement story. Minh restated a people-centred line: AI should serve people without replacing human authority, protect rights and privacy, and keep fairness, transparency, and accountability visible. Vietnam, he said, wants risk-proportionate rules that still leave room to test new solutions. The safety track explicitly calls for sharing risk-assessment methods and technical standards while respecting linguistic, cultural, and data-sovereignty differences.
Why it matters: Vietnamese firms already import foreign models while Hanoi drafts education and hub strategies. Regional testing partnerships can raise assurance—but only if Vietnamese evaluation labs and language data stay first-class participants.
What it means in practice
Vietnamese regulators and vendors should map which safety tests could be mutualised without surrendering datasets; confirm SME training seats under the skills track; assign an owner for ASEAN standards comments; run time-boxed reviews of foreign-model geopolitical bias risks; and prefer contracts that name Vietnamese audit exports. Anchor the proposal to Vietnam’s ASEAN AI hub strategy and national AI education framework drafts.
Caveats come first. Ministerial roundtables are not binding treaties; cooperation menus can stay rhetorical; and safety testing needs funded labs. AI4Vietnam therefore presents the three-track pitch as directional policy context until published workplans and pilot laboratories appear.
What to watch next: named Vietnamese institutes in joint testing; SME training quotas; and how Hanoi’s citywide AI education plan feeds the skills track. Readers can continue on the AI4Vietnam homepage, or browse the Newsroom for additional briefings.
Bottom line: treat this update as orientation, not instruction. Vietnamese AI diplomacy is seeking structured ASEAN–China channels and remains early. Organizations that benefit most will insist on measurable testing outputs, keep sovereignty clauses concrete, and refuse to confuse a roundtable speech with finished regional standards.