News
FSA and BOJ issue request on responding to frontier AI threats
2026-05-22
On May 22, the Financial Services Agency and the Bank of Japan jointly issued a request to financial institutions concerning "short-term responses by financial institutions in light of changes in threats posed by frontier AI." Reflecting the changing cyber threat landscape brought about by advanced AI, the request calls on financial institutions to strengthen their security responses in the short term.
Overview
On May 22, 2026, Japan's Financial Services Agency (FSA) and the Bank of Japan jointly issued a request to the heads of relevant institutions concerning "short-term responses by financial institutions in light of changes in threats posed by frontier AI." The letter was signed jointly by four senior FSA officials, including the directors-general of three bureaus, and an Executive Director of the Bank of Japan. It asks financial institutions to implement the short-term measures set out in an attachment under the direct involvement of senior management, including the chief executive. The attachment states that institutions are expected to make progress on the measures within roughly one month.
Background
The letter notes that cyber threats are rising as AI is used in attacks, dramatically increasing their speed and scale. So-called frontier AI is highly capable at discovering vulnerabilities and generating sophisticated attack code. As a result, vulnerabilities that were previously hard to find may be discovered in large numbers in a short time, and the interval between discovery and exploitation may shrink sharply. The FSA convened a public-private conference on April 24 and held the first meeting of a working-level group on May 14, which compiled the measures. The FSA also said it will proceed in line with "Project YATA-Shield," the government-wide package published on May 18 by the National Cybersecurity Office. At the same time, citing an evaluation report by the UK AI Security Institute (AISI), the document notes that frontier AI cannot currently be said to succeed in attacking well-defended IT systems. It therefore stresses that faithfully implementing basic measures under existing FSA guidelines remains important.
Key Points
The attachment lists nine responses to a large volume of vulnerabilities:
- Treat the response to frontier AI as a company-wide management issue.
- Identify the services and IT systems to prioritize, with externally facing systems such as internet banking at the top.
- Eliminate technical debt in those assets, for example by closing unnecessary network ports, removing privileged IDs and updating products that are no longer supported.
- Add human resources for patch application.
- Review maintenance contracts with vendors, including night and holiday response and service level agreements (SLAs) and objectives (SLOs).
- Make the patching process risk-based, prioritizing by the likelihood that an attack would succeed.
- Strengthen non-patch defenses, such as virtual patching via web application firewalls (WAF), multi-factor authentication and endpoint detection and response (EDR).
- Prepare for outages of priority services and systems by reviewing business continuity plans and criteria for proactive shutdowns.
- Maintain and strengthen external cooperation through Financials ISAC Japan and similar bodies.
On pre-patch testing, the document suggests weighing the risk of system failures from insufficient testing against the risk of attack from unpatched systems, and considering a reasonable reduction in testing where appropriate. For jointly operated systems and those provided by cloud providers, it asks institutions to clarify the division of responsibilities between users and providers and to confirm that contracts require appropriate reporting on patch status.
The document emphasizes that these are emergency measures, and that over the medium to long term institutions need to move toward automated vulnerability management.
Implications for Markets and Companies
For financial institutions, preparing for a large number of patches in a short period becomes a management-level issue. This is likely to require reviews of IT departments, vendor arrangements and maintenance contracts. Note that the threats are described as extending beyond in-house systems to third-party software, including open-source software, and cloud services. IT vendors serving financial institutions, as well as foreign-owned and Chinese-affiliated firms registered to conduct financial business in Japan, may also need to review their own arrangements.
Outlook and Timeline
The request is based on the current situation, and the document stresses that measures should be reviewed flexibly as AI developments evolve. Whether further measures will follow, and when, has not been indicated.
Comment from TAKARA
TAKARA continuously monitors the supervisory policies of authorities such as the FSA and the Bank of Japan, including requests related to cybersecurity. We provide information on procedural matters relating to registration as a financial instruments business operator and the subsequent development of internal controls.
This article provides general information based on published materials and news reports. It is not a solicitation or recommendation of any financial product and does not constitute investment advice.


