Katsuhiko Kato, chairman of the Japanese Bankers Association, said the industry may suspend some services, including ATMs, to counter AI-driven cyber threats.

This shift in strategy signals a move toward aggressive risk mitigation as financial institutions struggle to keep pace with the rapid evolution of malicious artificial intelligence. By prioritizing asset security over customer convenience, Japanese banks are acknowledging that traditional defense mechanisms may no longer be sufficient to prevent large-scale theft.

Kato said that he recognizes the dimension of threats has changed significantly [1]. He said that when choosing between customer convenience and the protection of vital assets, the priority is clear [1]. The association is now considering the temporary shutdown of specific digital and physical touchpoints if the risk of an AI-powered attack becomes too high.

These warnings follow a broader push for urgency within the Japanese financial sector. On May 22, 2026 [2], the Financial Services Agency and the Bank of Japan issued a joint statement requesting that financial institutions implement short-term responses to address the evolving threats posed by frontier AI [2].

While previous regulatory requests focused on general readiness and short-term responses, Kato's comments introduce the possibility of concrete service interruptions [1]. This represents a more severe level of intervention than the general guidelines previously discussed in industry webinars held earlier this month [3].

Kato said the priority is to reduce risk to protect customer assets from sophisticated attacks that leverage the latest AI capabilities [1]. The association has not yet specified the exact triggers that would lead to a service shutdown, but the possibility remains a key part of the current security framework.

“The dimension of threats has changed significantly”

The willingness to disable ATMs suggests that the 'frontier AI' threat is perceived as an existential risk to the integrity of the banking system rather than a manageable nuisance. This move indicates a transition from a 'defense-in-depth' strategy to a 'fail-safe' approach, where the only guaranteed way to protect assets during a high-intensity attack is to sever the connection to the outside world entirely.