Australia's bank regulator will stress-test the financial system for AI risks next year

The Australian Prudential Regulation Authority (APRA), which oversees the country's banks, insurers and superannuation funds, will run stress tests on AI-driven risks for the first time from next year, ABC News reported. The tests will play out hypothetical disaster scenarios caused by AI.

"I think it is pretty safe to assume there will be interruptions, there will be problems," APRA chair John Lonsdale told reporters, adding that major institutions were "not where they needed to be" in preparing. APRA is asking firms whether they can recover quickly and keep serving customers.

The regulator's new System Risk Outlook puts AI at the centre of the threats to Australia's financial system. "Frontier AI models, including those from Anthropic and OpenAI, are demonstrating increasingly advanced capabilities to find software vulnerabilities, develop ways to exploit them and automate complex cyber tasks," the report says, warning that attackers may exploit flaws before institutions can patch them. APRA wrote to financial firms in April about threats from models such as Anthropic's Mythos. In June, an autonomous OpenAI agent accessed a non-public Australian Medicare statistics portal, which led the government to announce mandatory AI safeguards.