State Bank of Vietnam introduces new regulations for high-risk artificial intelligence systems starting November 15

Executive Briefing
“The State Bank of Vietnam issued Circular 53 establishing automated transaction value...”
Key Takeaways
- 1.Circular 53/2026/TT-NHNN by the State Bank of Vietnam officially takes effect on November 15, 2026, regulating risk management for artificial intelligence systems in the banking sector.
- 2.Artificial intelligence systems automatically executing personal money transfers exceeding 10 million dong per transaction or totaling over 20 million dong daily are classified as high risk.
- 3.For institutional customers, the high-risk threshold applies to automated systems transferring over 50 million dong per transaction or exceeding 100 million dong in daily total transaction value.
- 4.The circular also classifies artificial intelligence systems automatically deciding credit granting for loans of 50 million dong or more into the high-risk group.
The State Bank of Vietnam issued Circular 53 establishing automated transaction value thresholds that classify artificial intelligence systems into high-risk categories. Banks must comply with strict control requirements and maintain human oversight mechanisms.
Key Developments and Technical Context
Circular 53/2026/TT-NHNN by the State Bank of Vietnam officially takes effect on November 15, 2026, regulating risk management for artificial intelligence systems in the banking sector.
Artificial intelligence systems automatically executing personal money transfers exceeding 10 million dong per transaction or totaling over 20 million dong daily are classified as high risk.
For institutional customers, the high-risk threshold applies to automated systems transferring over 50 million dong per transaction or exceeding 100 million dong in daily total transaction value.
The circular also classifies artificial intelligence systems automatically deciding credit granting for loans of 50 million dong or more into the high-risk group.
Strategic Analysis and Industry Implications
Establishing specific value thresholds helps regulators closely monitor large-scale financial automation activities, thereby mitigating security incidents and model discrepancies in the banking sector.
The mandatory requirement for human intervention mechanisms ensures that artificial intelligence technology merely acts as support and does not replace the legal authority of banking personnel.
Outlook and Key Factors to Watch
The specific level of detail for technical guidelines to evaluate cybersecurity and model controls for individual credit institutions was not clearly specified in the initial information.







