News
Australia's ASD Calls AI an 'Unprecedented' Cyber Threat, Says Disruptions Are Inevitable
In its annual report, the Australian Signals Directorate says AI was the defining force in the cyber threat landscape over the past year. Director-General Abigail Bradshaw says each increase in model capability will likely bring more disruption.
Contents
The Australian Signals Directorate (ASD), the country's signals intelligence agency, warned on Tuesday, 27 October, that artificial intelligence poses an "unprecedented" threat to cyber security. According to the agency, foreign governments are using AI to break into computer networks and critical infrastructure.
What the report says
The annual cyber threat report describes AI as "the defining force in the cyber threat landscape over the past year." According to the document, it lets attackers find vulnerabilities faster and exploit them at unprecedented scale, speed and sophistication. Euronews notes that ASD used some AI tools to compile the report itself.
According to ABC News, the number of incidents the agency responded to rose by 20 percent in the last financial year. The report also mentions threats that emerged in recent weeks from the unintended actions of AI agents and from model testing.
What is inevitable is that with each capability increase, we are likely to see more disruption. - Abigail Bradshaw, ASD director-general
Disruptions and new incidents
Bradshaw said the latest AI models have "redefined" the threat facing Australia. Disruption can mean taking down sites that need patching in order to secure them, as well as data breaches or service outages. She expects state actors to follow criminals in using AI to "fill their information gaps."
ABC notes that a few days earlier, government websites, including ASD's, were partially taken offline after vulnerabilities were found. Euronews places the PHP flaws, which led governments to shut down sites, in the same context.
ABC also mentions a swarm of AI agents blamed for breaching the security of Australian health data, described as the first autonomous hack of a government website. Euronews writes that in June an OpenAI model bypassed safeguards during training and hacked an Australian government website; the company only discovered it in August and informed the authorities in September.
ASD's offensive operations
The agency revealed previously secret details of its own operations. ASD operators got into the back end of the Amadey malware operation, which had reached 180,000 computers worldwide, and cut the controller off from its victims. The software was sold on the dark web for about $US500, and in one case it was found hidden in an illegitimate free version of Photoshop.
ASD also destroyed 5 terabytes of data stolen from a business that did not know it had been hacked. Among it were about 4.8 million files belonging to Australians.
AI on the defending side too
Tauri Gregory, ASD's acting deputy director-general of intelligence, stresses that AI widens the scope and scale of attacks but is also a major asset for defenders; the agency uses it to speed up its own software development. Bradshaw says more work is needed, especially in replacing legacy systems and learning to use AI defensively, which she called "a matter of urgency."
