AI-driven cyber threats surge in Australia, putting superannuation savings at risk
Advanced AI models have sparked an 80% jump in cyber attacks on Australian businesses and government, endangering the nation’s superannuation funds.
In its upcoming cyber-threat assessment, the Australian Signals Directorate highlighted that frontier AI technologies have accelerated and broadened cyber assaults, driving an 80% increase in incidents against Australian enterprises and public bodies. Models such as Anthropic’s Mythos are being used by both defenders and malicious actors to uncover vulnerabilities, exemplified by a recent PHP flaw that forced the Australian Bureau of Statistics and other sites offline for emergency patches.
Regulators warned that the nation’s superannuation system, holding more than $4.5 trillion across 25 million accounts, could face massive withdrawals if AI-enhanced attacks succeed. During the 2025-26 financial year the ASD issued over 4,400 alerts, handled more than 1,500 incidents, blocked access to 564 million malicious domains and responded to 139 ransomware cases. The average cost of attacks rose 16% to $93,500, with medium-sized firms seeing a 45% jump to $141,300.
Authorities stressed the need for rapid cyber-hygiene upgrades and the removal of legacy software from the internet to limit exploitable entry points. Defence Minister Richard Marles echoed the call, urging businesses to prioritise security over service continuity as AI reshapes the threat landscape.
Why it matters
AI-enhanced cyber attacks could compromise billions in retirement savings and destabilise Australia's economy.
In this story
