APRA Warns Financial Firms on AI Cyber Gaps

Australia's financial regulator has warned banks, insurers, and superannuation funds that their governance and cybersecurity practices are not keeping pace with rapid advances in artificial intelligence, particularly frontier models like Anthropic's Claude Mythos that could enable faster and more devastating cyberattacks.

In a letter sent this week to regulated entities, the Australian Prudential Regulation Authority (APRA) outlined findings from a supervisory review conducted late last year. The review assessed AI deployment across major institutions and identified gaps in risk management, operational resilience, and board oversight. APRA highlighted how AI is accelerating the discovery and exploitation of vulnerabilities, allowing threat actors to increase the probability, speed, and scale of attacks on financial systems.

"Frontier AI models such as Anthropics Claude Mythos, which could enhance the discovery of vulnerabilities by bad actors, are expected to further increase the probability, speed and scale of cyber attacks," the letter stated. APRA noted that many organizations treat AI risks like traditional technology threats, overlooking unique challenges such as predictive...