Top cyber spy braces for more AI-powered hacking attacks

The Australian Signals Directorate (ASD) is bracing for a surge in AI-powered cyber attacks, warning the country to prepare for more frequent disruptions to everyday services. AI is accelerating the number and severity of cyber threats from criminal networks and state-aligned operators, according to the agency. ASD director-general Abigail Bradshaw states that AI frontier models have fundamentally altered the threat landscape facing Australia, with disruptions potentially manifesting as data breaches, service outages, or other forms of digital network disruption.

The warning follows a recent partial shutdown of government websites following the discovery of vulnerabilities, and a 20% increase in cyber security incidents reported last financial year. The release of frontier models like Anthropic's Mythos, touted for their vulnerability-finding capabilities, has further heightened concerns.

In response, the ASD has established protocols to act as an early warning system with AI companies, including OpenAI, and has collaborated on offensive operations against criminal groups such as Amadey, which was found exploiting 180,000 compromised computers worldwide, including in Australia. AI is increasingly being used both to enhance cyber defence capabilities and to facilitate attacks, with the agency leveraging AI to develop new tools and capabilities more rapidly.