AI Governance Policy

MirAI researches, develops and integrates AI Agents into the systems of banks and financial institutions, where every decision is tied to the flow of money and to customer trust. We believe AI serves people and does not replace human authority or responsibility. This Policy sets out the principles MirAI follows throughout the life cycle of every AI system, from design and development to deployment and operation.

This Policy is based on Vietnam's Law on Artificial Intelligence (Law No. 134/2025/QH15), Japan's AI Guidelines for Business (AI事業者ガイドライン) issued by the Ministry of Internal Affairs and Communications and the Ministry of Economy, Trade and Industry, and the financial-sector regulations of each country where our customers operate.

1.Humans keep the final decision

Decisions that directly affect a bank's customers, such as credit approval, blocking a transaction or refusing a service, are always approved by a person or open to human intervention. Every AI Agent has a mechanism that lets a person pause or fully stop it when needed.

2.Safety and security

An AI Agent is given only the minimum access its task requires. Before going live, each Agent is tested in an isolated environment, assessed for risk and fitted with guardrails matched to the sensitivity of the work.

3.Data protection and privacy

Customer data entrusted to us is classified by sensitivity and stored and processed in line with the customer's data-location requirements. MirAI does not use customer data to train models without the customer's written consent.

4.Compliance with laws and regulations

Every AI system is designed to comply with the laws of Vietnam, Japan and the country where the customer operates, as well as the customer's own internal rules.

5.Transparency and accountability

Every action of an AI Agent is logged for audit. Results produced by AI can be explained to a degree appropriate to the work. Content generated by AI is clearly identified as required by regulation.

6.Fairness

For models that concern individual customers, especially credit scoring, we test for and reduce bias so that no unreasonable discrimination occurs.

7.Risk management and continuous improvement

Risks are assessed from the design stage, monitored throughout operation and handled promptly when incidents occur. This Policy is reviewed regularly and updated as technology and law develop. Every MirAI member is trained in responsible AI and follows this Policy.