Financial System Report Annex "Use and Risk Management of Generative AI by Japanese Financial Institutions"

In a report on the use and risk management of Generative AI by Japanese financial institutions, the Bank of Japan has found that GenAI is rapidly expanding from administrative tasks to core operations. This expansion is anticipated to continue as technology advances. Financial institutions must understand the unique risks associated with GenAI, such as information leakage and unpredictable outputs, and strengthen their risk management frameworks using a risk-based approach.

The Bank of Japan recently surveyed 150 financial institutions, marking the third such survey since fiscal year 2024. This report presents the findings from interviews with IT vendors and financial institutions, detailing the current state and challenges of GenAI use and risk management in the financial sector.

Survey results indicate that Japanese financial institutions plan to further develop their use of GenAI, including in core operations. As a result, the complexities and diversity of risks are likely to grow. Specific areas requiring careful risk management and governance include presenting AI-generated outputs to customers, deploying autonomous AI agents, and integrating cutting-edge frontier AI.

To responsibly employ GenAI, including frontier AI, while ensuring stable business operations, senior management must recognize both the benefits and risks of adopting new technologies. Additionally, they must actively participate in securing resources and enhancing frameworks for effective risk management.

The Bank of Japan will maintain dialogue with financial institutions on GenAI use and risk management via on-site examinations, monitoring, and seminars. For permission to reproduce or copy the report's content, contact the Financial System and Bank Examination Department via email. When quoting, reproducing, or copying the report for non-commercial purposes, credit the source.