Finance leaders of the Group of Seven have agreed to run tabletop exercises that simulate cyberattacks powered by frontier AI, Japan's Finance Minister Satsuki Katayama said on Thursday, in the most concrete preparedness step yet by major economies against a threat regulators have mostly discussed in warnings rather than rehearsals.
Katayama made the announcement after a meeting of G7 finance ministers and central bank governors held on the sidelines of the IMF and World Bank annual gatherings in Bangkok. "The G7 finance deputies had been discussing the idea of doing tabletop exercises on frontier AI, which was signed off at the ministerial level," she told reporters.
According to a Japanese finance ministry official who attended the meeting, the simulations will be developed by G7 finance authorities themselves to test how governments and central banks would respond to an AI-driven attack on financial infrastructure. The results will be used to judge whether existing response protocols hold up, and the manuals that govern incident response will be rewritten where they fall short. Depending on the scenario, private financial institutions could take part. No timetable or scenario detail has been published.
The agreement lands amid a run of alarm from the institutions that supervise the global financial system. The International Monetary Fund warned this week that public confidence could be undermined by an AI-enabled cyberattack on a critical service of a financial institution. The Bank of England, meanwhile, is working with banks on what its deputy governor described as deep cyber-recovery options, saying lenders can no longer assume their defenses are immune to AI-enabled attacks.
The concern is not hypothetical capability but speed. Security researchers and vendors have documented AI compressing the gap between vulnerability discovery and weaponization to under a day, and regulators have begun citing rogue autonomous agents as a potential source of instability rather than a science-fiction scenario. For a payment rail or a settlement system, an attacker that operates at machine speed compresses the time available for human response to nearly zero.
A tabletop exercise does not fix any of that, and skeptics of voluntary coordination note that drills are easy to announce and hard to grade. But the design choices are telling: the exercises are meant to stress real protocols and produce revisions to response manuals, and the possible inclusion of private banks would turn a G7 scenario into a de facto benchmark for what supervisors expect recovery plans to look like.
For financial institutions, the underlying message cuts both ways. AI is already their best defensive tool for triaging alerts and reconstructing intrusions, and it is the thing their adversaries are using to get faster. The G7's bet is that the gap between the two can be managed — provided someone has rehearsed it before it happens for real.
Comments (0)
Log in to join the discussion
Log InNo comments yet