Story
Oct 17, 2026
G7 Banks Brace for AI Attacks That Could Move Faster Than Their Defenses
G7 policymakers see AI as a force that can strengthen cyber defenses but also sharply magnify the threat to the financial system. Their answer is to test recovery and coordination before an attack on a major bank or critical infrastructure turns into a wider crisis.
The warning signs came first. On Oct. 13, the International Monetary Fund said frontier AI could speed up both cyberattacks and defenses, making resilience, incident reporting and crisis planning more urgent. It noted that attackers have increasingly exploited vulnerabilities before software providers can patch them — a reversal of the breathing room defenders once had.1
Two days later, the IMF sharpened the concern: as finance becomes more digital, AI is compressing the contest between attacker and defender “to machine speed.” A serious breach affecting payments, trading or access to deposits, it said, could produce immediate losses and erode confidence.1
Against that backdrop, G7 finance deputies agreed to run tabletop exercises simulating AI-powered attacks on banks, Japanese Finance Minister Satsuki Katayama said Thursday after meetings alongside the IMF’s annual gathering in Bangkok. The ministerial-level sign-off means finance ministries and central banks will test how they would respond; private financial institutions could also be involved.1
The initiative reflects a shift from trying to prevent every intrusion to preparing for the moment defenses fail. Bank of England Deputy Governor Sarah Breeden said banks must plan how to recover from AI-enabled breaches rather than assume attacks will be stopped at the gate. “Cyber is always the thing that I say keeps me up at night,” she said, warning that a systemically important bank or core infrastructure could be attacked and left “paralyzed.”1
Japan’s finance minister separately confirmed Thursday that the G7 drills are intended to address growing alarm over AI’s capacity to supercharge hacking.2
The common message is blunt: an AI-enhanced cyber shock is no longer being treated as a distant technical risk, but as a potential test of financial stability itself.