

Australia’s latest national threat assessment confirms what SOC teams running flat out on every shift already know: attacks now move faster than the operating model built to stop them. In its Annual Cyber Threat Report 2025–26, the Australian Signals Directorate’s Australian Cyber Security Centre (ASD’s ACSC) says it responded to more than 1,500 cyber security incidents, up more than 20%. It also notified organisations of potential malicious activity more than 4,400 times, an increase of almost 80%.
The report calls AI “the defining force in the cyber threat landscape” this year. Attackers are using it to find and exploit vulnerabilities faster and at greater scale. ASD also flags a newer risk: harm caused by the unintended actions of AI agents and model testing.
For CISOs, that pairing is the real story. AI is speeding up the adversary while adding new risk inside the defender’s own stack.
Three findings stand out, and every security leader should be across them.
Anomali’s own country-level research adds detail. In our report, Australia’s Ransomware Threat Landscape, we tracked 73 distinct ransomware entities targeting Australian industries during 2025. Nine of the 18 industries we analysed were in scope for at least half of them. Technology topped the list, targeted by 67 entities (92%), followed by financial services at 66 (90%).
Breadth matters as much as volume. Our threat researchers observed The Play group across 12 of 18 industries, and the largest share of entities operated across seven or eight sectors. The tradecraft is familiar: valid accounts, standard admin tools, data theft before encryption, and deliberate tampering with logs and recovery. Our 2026 forecast expects that sequence to compress further, from breach to impact in days rather than weeks.
Put those findings together and the problem isn’t visibility. Most SOCs see plenty. The problem is what happens between an alert and a decision.
In the old operating model, intelligence and telemetry live in separate tools and meet manually. An alert fires in the SIEM, an analyst queries the threat intelligence platform, and context is applied case by case. That worked when attackers moved at human speed. It fails when an affiliate can go from a stolen VPN credential to encrypted servers in days.
The tempting response is to point AI at the problem and assume she’ll be right. But an agent given raw, fragmented, unenriched telemetry inherits every flaw in that data. It makes wrong decisions faster than the analyst it was meant to help. ASD’s warning about unintended agent actions makes the same point: speed without governance is just a new kind of risk.
The old model produced alerts. The new operating model has to produce governed decisions…at machine speed.
Building that model starts with the data, not the agent. Fidelity means three things happen before any analyst or agent sees an event:
For SOC leaders, the payoff is a change in the question your team asks. When alerts arrive already attributed, mapped to MITRE ATT&CK and scored, analysts stop asking what an alert is and start asking what to do about it. That matters when a nation-state operator and a ransomware affiliate can look alike at first glance.
For CISOs, the bar is higher than speed. Every automated action should be explainable, auditable and reversible, with humans in the loop for consequential decisions. That is the standard for a decision you can defend to your board, your regulator and your night shift.
This is the thinking behind Anomali’s Agentic SOC and the Intelligent Unification Layer underneath it: unify data, ground decisions in operational intelligence, govern by policy, and execute at machine speed.
ASD closes by asking Australians to turn cyber awareness into practical action. For security leaders, that means going beyond detection to action, with governed decisions at machine speed.
Next step: Download Anomali’s Australia’s Ransomware Threat Landscape report for the full industry targeting data, entity profiles and 2026 forecast: https://www.anomali.com/resources/datasheets/australias-ransomware-threat-landscape
FEATURED RESOURCES

