AI speeds up cyber attacks: Australia urges organisations to patch faster
Australia’s ASD says the incidents it handled rose 20% in its annual report published on 26 October. It warns that AI may shorten the path from vulnerability discovery to exploitation, with implications for network defence and essential services.
Translated from the French original. In case of discrepancy, the French version prevails.
AI does not necessarily create a new class of cyberattack, but it can accelerate familiar methods. That is the assessment highlighted by the Australian Signals Directorate (ASD) in its 2025–26 annual report, published on 26 October and presented by the agency on 27 October. [1][2]
More incidents, more notifications
For 2025–26, the Australian Cyber Security Centre says it responded to over 1,500 incidents, up 20% year-on-year and 36% in five years. It issued over 4,400 notifications of potentially malicious activity, nearly 80% more, and received over 82,000 cybercrime reports. [1]
These figures describe activity known to the agency and reports it received; they are not a complete count of every attack in Australia. The report says much cybercrime may go unreported. [1]
The exploitation window is narrowing
ASD says AI can help malicious actors analyse public information, select targets and identify or chain vulnerabilities. The report notes that some flaws have been exploited within 48 hours of a patch or mitigation notice. ASD expects AI may compress that window further; this is the agency’s assessment, not an independent measure of every attack. [1]
The report also points to defensive uses: earlier vulnerability detection, monitoring unusual behaviour and faster incident response. The challenge is organisations’ ability to adapt as well as model capability. [1]
Resilience is a continuity issue
The networks at risk include businesses, critical infrastructure and individuals, not only government. The civil-defence link is practical: the same digital networks support economic activity, essential services and public or military functions. This does not automatically make every technology mentioned a legally controlled dual-use item. [1]
Australia urges organisations to patch quickly, log events, manage supply-chain risks and prepare for service disruption. It says security may need to take priority over availability. [2]
A warning, not proof of autonomous attacks
ABC reports that ASD director-general Abigail Bradshaw expects more frequent disruptions and that the agency has established early-warning protocols with AI companies. These statements are attributed to ASD and ABC; they do not prove that incidents in the report were carried out end-to-end by autonomous agents. [3]
Canberra’s warning is a race against time: detect earlier, patch faster and prepare for temporary service interruptions to prevent more serious compromise.
Fact-check
The fact-check, point by point
Each claim, checked against public documents. Click on a point to read the full analysis.
What the official version says
Australia’s Signals Directorate published its 2025–26 cyber report on 26 October. It reports over 1,500 incidents handled and more than 4,400 notifications. The agency says AI can accelerate some attack stages but may also support defenders.
Claims checked
Confirmed by the report ASD responded to over 1,500 incidents in 2025–26.
This is the number handled by the agency, not a complete count of attacks affecting Australia.[1]
Not established AI caused the increase in incidents.
False AI only helps attackers.
The report also identifies potential defensive uses.[1]
Sourced counter-arguments
ASD has distinct visibility into reports and response operations, but its figures do not measure unknown or unreported incidents. Its AI assessment is a threat evaluation, not a complete causal attribution.[1]
Questions that remain open
- What share of incidents involved AI?
- How is time-to-exploit measured?
- Have early-warning protocols changed security practice?
In short
The report highlights an operational constraint: attackers may accelerate parts of the process, while defenders depend on organisations’ ability to detect, patch and maintain services.
Sources
- Annual Cyber Threat Report 2025–26 (published 26 October 2026) — Australian Signals Directorate / Australian Cyber Security Centre · accessed 27 October 2026
- ASD releases seventh annual Cyber Threat Report 2025-26 (27 October 2026) — Australian Signals Directorate · accessed 27 October 2026
- Disruptions 'inevitable' as AI-powered cyber threats grow (27 October 2026) — ABC News · accessed 27 October 2026
AI-assisted content: this article was written with the help of artificial intelligence from public sources, then reviewed and approved by a human editor before publication. Spotted an error, want to respond? Exercise your right of reply · our method.
