Finance leaders from the G7 countries have agreed to hold joint tabletop exercises to practise responding to AI-enabled cyber attacks on the financial system. Why it matters: finance is one of the first sectors where an AI-driven attack could spread faster than the institutions trying to stop it, and the G7 is now rehearsing for that scenario instead of just warning about it.
October 16, 2026
What happened?
Japanese Finance Minister Satsuki Katayama told reporters on Thursday that G7 finance leaders had agreed to run tabletop exercises aimed at AI cyber attacks. The agreement followed a meeting of G7 finance ministers and central bank governors held in Bangkok, on the sidelines of the International Monetary Fund’s annual gathering.
Tabletop exercises are simulations rather than live drills. Officials work through a made-up incident step by step — what gets hit, who gets called, what gets shut down, and in what order — to find gaps in the plan before a real attack tests it.
What will the exercises actually test?
According to a Japanese finance ministry official who attended the meeting, G7 finance authorities will develop the simulations themselves. The results will be used to check whether existing response protocols still work against AI-driven attacks and to revise the manuals officials rely on during an incident.
The exercises are being run by finance ministries and central banks, and private financial institutions could also take part, depending on the kind of simulation being run. No dates for the exercises were announced in the reporting.
Why is this happening now?
The agreement follows an IMF warning this week that an AI-enabled cyber attack on a critical service at a financial institution could undermine confidence in the system. That concern sits alongside a wider regulatory push: banking watchdogs in several countries have been stepping up work on AI risks, including attacks on banks and the possibility of rogue AI agents destabilising finance.
The Bank of England offered a parallel example on Thursday. Its deputy governor said the bank is working with lenders on deeper cyber-recovery options, and that banks can no longer assume their defences are immune to AI-enabled attacks.
Why it matters
For readers, the practical takeaway is about preparation, not panic. The G7 move does not change any rule for banks or AI companies today, and no exercise schedule has been published. What it does show is that finance officials now treat an AI-powered attack as a scenario worth rehearsing at the highest level — the same week global financial bodies were already warning banks to keep clear human control over AI agents. If your money sits with a large bank, expect more questions from regulators about how that bank would recover, not just how it would defend.
Frequently asked questions
What is a tabletop exercise?
It is a discussion-based simulation. Participants talk through a fictional incident and their planned responses, so weaknesses in protocols and communication can be found without running a live test on real systems.
Who agreed to the G7 exercises?
G7 finance ministers and central bank governors, meeting in Bangkok alongside the IMF’s annual gathering. Japanese Finance Minister Satsuki Katayama announced the agreement to reporters on Thursday.
Will private banks take part?
Possibly. A Japanese finance ministry official said private financial institutions could be included, depending on the nature of each simulation.
When will the exercises happen?
No dates were announced in the reporting. The stated next step is for G7 finance authorities to develop the simulations and then use the results to update their response manuals.
Sources: Reuters