Reading Vietnam's AI Law as a Financial Institution

Vietnam's AI Law, Law No. 134/2025/QH15, has been in force since March 2026, and the finance sector's compliance date is September 2027. What a bank or insurer should be doing in the meantime.

Vietnam brought its Artificial Intelligence Law into operation on 1 March 2026, the first jurisdiction anywhere to do so in full, ahead of the European Union's phased timetable. For a financial institution operating in the country, the more consequential date is the sector compliance deadline of 1 September 2027.

What already applies The law binds domestic and foreign developers, providers and deployers operating in Vietnam. Its most immediate obligations are disclosure obligations: labelling AI-generated content, and making it clear when a person is interacting with an automated system rather than a human. These are not hypothetical for an institution running automated customer service or generating marketing content at scale.

What the runway is for Between now and September 2027, an institution needs to be able to demonstrate governance rather than assert it. In practice that means a register of AI systems, a risk classification with stated criteria, impact assessments for customer-facing decisions, and an operating model that names who owns each of those things.

Where the gap usually sits Not in technology awareness. Institutions in Vietnam generally know what AI they run and roughly what it does. The gap is documentation: the evidence trail that lets a supervisor, an external auditor or an internal control function reconstruct why a system was approved, on what basis, and under whose authority.

Conclusion Eighteen months sounds generous until it has been spent deciding. The institutions that will be comfortable in September 2027 are the ones that started with a register and a gap analysis, not the ones that started by buying a platform.