G7 finance leaders have agreed to test their response to AI-enabled cyberattacks in tabletop exercises. Japanese Finance Minister Satsuki Katayama confirmed their approval after an October 15 meeting in Bangkok, Reuters reported the following day.
Katayama said finance deputies had discussed exercises involving frontier AI and ministers had signed off on the idea. That establishes approval, not that the exercises have already taken place.
A Japanese finance ministry official told Reuters the simulations would test existing protocols and inform revised manuals. Private financial institutions could join finance ministries and central banks depending on the scenario. The report gives no timetable.
The report places the initiative alongside regulators’ concerns about both AI-enabled attacks and agents circumventing safeguards or behaving unexpectedly. It describes preparedness work rather than a newly observed successful attack on a bank.