G7 finance chiefs agree to rehearse AI-driven cyber attacks on the financial system, Reuters reports

Finance ministers and central bank governors of the Group of Seven have agreed to hold tabletop exercises that simulate AI-driven cyber attacks on the financial system, Reuters reported on 16 October. They met on the sidelines of the IMF's annual meetings in Bangkok.

Japan's finance minister, Satsuki Katayama, told reporters on 15 October that G7 finance deputies had been discussing “tabletop exercises on frontier AI, which was signed off at the ministerial level”, according to Reuters. A Japanese finance ministry official told Reuters that the simulations would test how authorities respond to AI-driven attacks, that the results would be used to check protocols and revise manuals, and that private financial institutions could take part depending on the simulation.

No start date was given, and no G7 communiqué or statement from Japan's finance ministry or the US Treasury on the exercises could be found. Reuters says the agreement comes after an IMF warning this week that an AI-enabled attack on a critical financial service could undermine confidence.

It is the latest sign that financial supervisors are treating frontier AI as a cyber risk to the system as a whole. On 14 October the Switchboard reported that US bank regulators were questioning lenders about rogue AI agents, and on 16 October that the Financial Stability Board had finalised 12 sound practices for AI in finance.

• Reported Reuters reports that G7 finance ministers and central bank governors agreed to hold tabletop exercises simulating AI-driven cyber attacks, signed off at ministerial level according to Japan's finance minister. Reuters, via The Star
• Reported A Japanese official told Reuters private financial institutions could take part; no start date was given. Reuters, via TimesLIVE