
Hoang Minh Tien, deputy director of the Information Technology Department under the State Bank of Vietnam, said AI would only be allowed to automatically process transactions within specified limits under the central bank's proposed regulations. Photo: Supplied
The State Bank of Vietnam (SBV), known as the central bank, is preparing a draft circular on safety, risk management, and conditions for deploying AI in banking operations, Hoang Minh Tien, deputy director of the SBV’s Information Technology Department, said at a seminar on Wednesday.
The move comes as banks increasingly use AI for credit scoring, loan assessments, fraud detection, anti-money laundering, customer service, and electronic identity verification.
Under the draft, AI systems would be classified into three categories based on risk levels: high, medium, and low.
Two applications would be considered high-risk: systems that automatically execute electronic transactions and those that independently make lending decisions.
For payment services, AI would be allowed to automatically process transactions ranging from VND10 million ($380) to less than VND100 million ($3,800) for individual customers and from VND50 million ($1,900) to less than VND500 million ($19,000) for corporate clients.
“AI will only be permitted to make autonomous decisions within these thresholds. Transactions exceeding these limits will require human intervention and approval,” Tien said.
The draft also proposes limits on AI-driven credit approvals.
Loans approved automatically by AI would be classified as high-risk when their value ranges from VND50 million ($1,900) to less than VND200 million ($7,600) at people’s credit funds and from VND50 million ($1,900) to less than VND400 million ($15,200) at other credit institutions.
According to SBV officials, while AI is becoming a core technology for improving productivity, streamlining operations, and personalizing services, it also creates risks related to data quality, cybersecurity, model performance, and accountability.
The central bank said the proposed regulations are intended to ensure AI is deployed safely and responsibly while maintaining human oversight over decisions that could significantly affect customers.
Max: 1500 characters
There are no comments yet. Be the first to comment.