Australia’s cyber spy chief warns AI will fuel disruption

Australia’s Signals Directorate has warned that artificial intelligence will fuel more disruptive cyberattacks and help attackers locate and exploit vulnerabilities at speed and scale. In its annual cyber threat report, ASD director-general Abigail Bradshaw says advanced AI models are redefining the threat landscape, with disruption potentially manifesting as patching-driven website takedowns, data breaches or service outages.

The warning follows recent outages on government sites tied to vulnerabilities, and coverage noting hackers leveraged AI-enabled methods to discover flaws.

The report paints a broad picture: state-backed and criminal actors continue to pursue sensitive information across government networks, critical infrastructure, and private sector systems. Figures cited diverge depending on the metric: ABC reports a 20% rise in incidents ASD handled in the last financial year, while AFP tallies more than 4,400 warnings of potentially malicious activity—an 80% increase from the prior year.

The ASD also described offensive work, including taking down Amadey, a malware network found in a free Photoshop copy with thousands of infected machines worldwide, and disrupting a data sale involving 5 terabytes of stolen business data.

The agency emphasises AI as both a threat and a defence tool, noting that frontier models like Anthropic’s Mythos have shifted risk dynamics, while ASD teams are incorporating AI to speed up defensive capabilities and require closer collaboration with AI labs and Five Eyes partners before testing new models.