Summary:

The Australian Signals Directorate's Australian Cyber Security Centre (ASD's ACSC) released its Annual Cyber Threat Report 2025–26 on 27 October 2026. The report identifies artificial intelligence (AI) as the defining force in the cyber threat landscape over the past year, with a new generation of frontier models making tools more accessible and lowering the effort required to target Australian critical infrastructure (CI), governments, businesses and individuals. The question, the report states, is no longer whether organisations will face cyber threats, but how prepared they are to respond.

In FY2025–26 ASD's ACSC responded to more than 1,500 cyber security incidents, an increase of more than 20% on the previous year and 36% over five years, and notified entities more than 4,400 times of potentially malicious activity, an increase of almost 80%. Beneath the volume figures the severity rose: extensive compromises increased from 26 to 40, extensive ransomware incidents from 16 to 24, and extensive compromises arising from distributed denial of service activity from one to seven. Only one incident was recorded at C2. ASD's ACSC categorises each incident it responds to from C1, the most severe, to C6, the least severe, based on severity of impact and the significance of the organisation's impact to Australia.

Behind those figures, the report describes two categories of actor using the same tools. State-sponsored actors are seeking to pre-position within Australian networks for intelligence collection or disruption at a time of their choosing, while cybercriminals operate through a service-based market that AI has made faster and cheaper to enter. The common effect is compression: ASD's ACSC has observed exploitation of some vulnerabilities within 48 hours of a patch being released, and expects AI-driven tools to narrow that window further.

Detailed Report

1. The Strategic Picture

The report opens by stating that technology is evolving at accelerated rates and continues to affect cyber security in both positive and negative ways. Throughout FY2025–26, AI tools became more accessible, driven by a new generation of frontier AI models which, in its words, “have changed the game with speed and opportunity, but they have also generated significant uncertainty.”

As those tools become more accessible, malicious cyber actors require less effort to target Australian CI, governments, businesses and individuals, whether for financial gain, espionage, or to pre-position on systems for later disruption. Both state-sponsored actors and cybercriminals can use AI to identify vulnerabilities, develop exploits and launch attacks at greater scale than previously possible.

State-sponsored cyber actors continued to target Australian organisations primarily to gain access to sensitive or non-public information, activity that extends beyond government networks to CI entities, private businesses and individuals. As the global security environment deteriorates, ASD assesses that these actors are increasingly seeking to pre-position within networks to enable future intelligence collection, disruption, or access to higher-value targets “at a time of their choosing,” with some motivated to maintain that access covertly in order to disrupt essential services during crisis or conflict. The trend is not confined to Australia and is evident across multiple global conflicts currently underway. Cyber, the report concludes, is “a normalised and highly integrated feature of conventional warfare.”

The distinction between state-sponsored and cybercriminal activity continues to blur. Despite differing objectives, both often exploit the same vulnerabilities and use similar tactics, techniques and procedures (TTPs) to compromise systems and extract data, including the use of AI.

The report cites the 2026 National Defence Strategy, which identifies deteriorating trends in Australia's strategic environment as having broadened and intensified, with the international system under greater strain than at any time since the end of the Cold War. Authoritarian regimes are described as becoming more willing and able to disrupt CI, conduct cyber intrusions and engage in misinformation and disinformation campaigns.

2. The Year in Numbers

Incidents and Notifications

ASD's ACSC responded to more than 1,500 cyber security incidents in FY2025–26, an increase of more than 20% from the previous year and a 36% rise over five years. Of these, 362 were identified following proactive notifications to entities about potential malicious activity on their networks.

The directorate issued more than 4,400 notifications of suspicious cyber activity, almost 80% more than in FY2024–25. This comprised more than 2,200 individual notifications, up 33%, and collective notifications issued to more than 2,100 entities. More than 51% of individual notifications originated from declassified intelligence. A single collective notification to 340 entities in May 2026, concerning common vulnerabilities and exposures (CVEs) affecting cPanel, resulted in more than 60 confirmed incidents.

Severity

Incidents categorised at C3 or above often involve federal and state governments, large organisations, academia and supply chains. While the number of higher-categorised incidents decreased in FY2025–26, with only one C2 incident recorded, extensive compromises increased from 26 in FY2024–25 to 40 in FY2025–26.

Denial of service (DoS) and distributed denial of service (DDoS) incident numbers fell sharply. ASD's ACSC responded to more than 80 such incidents, down more than 60% on the previous year, a fall attributed to the absence of any coordinated DDoS campaigns during the period. Those incidents that did occur produced more significant impacts: extensive compromises arising from DDoS activity rose from one in FY2024–25 to seven in FY2025–26.

Ransomware and data breaches

The same pattern appears in ransomware. ASD's ACSC responded to 139 ransomware incidents, consistent with the previous year, and the three Australian and New Zealand Standard Industrial Classification (ANZSIC) divisions most affected were unchanged: Health Care, Professional, Scientific and Technical Services, and Manufacturing. Severity rose, with extensive ransomware incidents increasing from 16 to 24.

Data breaches accounted for 98 incidents, more than 6% of the total and an increase of 66% on the previous year. Publicly reported CVEs increased 30%. The average self-reported cost of cybercrime was $58,130, with identity fraud, online shopping fraud and online banking fraud the most frequently reported types. Identity fraud remained the most reported cybercrime, down 18% on the previous year.

Separately, the Office of the Australian Information Commissioner recorded 1,205 notifications under its Notifiable Data Breaches scheme in 2025, the highest number in a year since the scheme commenced in 2018. Between July and December 2025, malicious or criminal attacks accounted for 405 of those breaches, the largest single source for the period. The health sector reported 64 notifications of this kind over the six months.

Critical infrastructure and reporting sectors

CI incidents accounted for 10% of all incidents, similar to the previous year. ASD's ACSC responded to 154 CI incidents, a decrease of almost 7%. The three ANZSIC divisions reporting the most CI incidents were Financial and Insurance Services at 30%, Public Administration and Safety at 21%, and Electricity, Gas, Water and Waste Services at 17%.

Government again reported the highest number of incidents, with the Federal Government at 32% and state and local government at 16%, a concentration the report attributes to additional reporting obligations. Health Care and Social Assistance, at 7%, rose to become the most frequently reporting non-government sector. Manufacturing entered the top ten on the largest increase in incident reports of any sector, while Transport dropped out.

3. The Threat Actors

State-sponsored actors

Great power competition and escalating geopolitical tensions are reflected in cyberspace, and over FY2025–26 Australia continued to experience persistent reconnaissance and testing of government and privately owned networks by state-backed actors.

These actors have targeted Australian organisations primarily to access sensitive or non-public information, pursuing material that fills their own intelligence gaps and reduces their tactical, operational, strategic and commercial risk. Such information is not exclusively held by governments and can, ASD assesses, almost certainly be gleaned from the networks of CI entities, private businesses and individuals.

In the Indo-Pacific, ASD assesses that state-sponsored actors are possibly preparing for eventualities such as crisis or conflict by pre-positioning capabilities for disruptive effects, or to access higher-value target networks for use at a time of their choosing. Disruption can serve strategic goals such as limiting the ability to respond to an event or causing social unrest. These actors use low-effort or low-cost techniques for opportunistic attacks, reserve advanced capabilities for high-value targets, and are investing in disruptive capabilities for future use in crisis or conflict.

The report identifies five state-sponsored actors of concern:

•       Advanced Persistent Threat 40 (APT40) — People’s Republic of China (PRC) state-sponsored, active since 2009, targeting Australian networks

•       Lazarus Group — Democratic People’s Republic of Korea (DPRK) state-sponsored, active since 2009, targeting CI worldwide

•       Salt Typhoon — PRC state-sponsored, active since 2019, targeting 200 organisations across 80 countries

•       Volt Typhoon — PRC state-sponsored, active since 2021, targeting CI across the United States and Asia Pacific

•       Advanced Persistent Threat 29 (APT29) — linked to Russian intelligence, active since 2008, targeting CI and Western organisations

Effective cyber attacks on vital operational technology (OT) and enabling systems, the report states, “could incapacitate Australia.”

In April 2026, ASD's ACSC joined international partners in highlighting a major shift in the TTPs used by China-nexus cyber actors: a move away from individually procured infrastructure towards large-scale networks of compromised devices, primarily small office/home office routers alongside Internet of Things (IoT) and smart devices. Most China-nexus threat actors are believed to be using these covert networks, which allow them to operate in a low-cost, deniable way that disguises the origin and attribution of their activity.

Cybercriminals

Cybercrime remains persistent and damaging across all facets of Australia's economy, with digital infrastructure and the information companies hold being prime targets. ASD's ACSC received more than 82,000 cybercrime reports through ReportCyber during the financial year, though it is likely most cybercrime goes unreported.

Ransomware is described as one of the most disruptive, costly and persistent cyber challenges facing Australia, having evolved over the past decade from basic encryption and extortion into an interconnected ecosystem operating through borderless online spaces that complicate investigation. The growth of the Cybercrime-as-a-Service (CaaS) model has made cybercrime more accessible and scalable, allowing less skilled actors to acquire ready-made malware and infrastructure; under the Ransomware-as-a-Service (RaaS) variant, affiliates purchase and customise ransomware, then share a percentage of profits with the developer. Paying a ransom, ASD states, does not guarantee recovery of data, nor prevent it from being sold or leaked online, and helps promote ransomware as a profitable criminal enterprise.

Hacktivists

Hacktivists are assessed as increasingly likely to pose a disruptive threat to government and private sectors during crisis or conflict. Though non-state-sponsored, they can be patriotically motivated, targeting those they perceive as threats to their own nation's geopolitical objectives. Australia's position in the Indo-Pacific places it at the forefront of many emerging global issues, and its reliance on international supply chains means hacktivist disruption can affect Australia even when it is not the direct target.

Two examples illustrate the point. On 11 March 2026, Iranian hackers claimed responsibility for compromising Stryker, a United States-based company that regularly imports surgical and pharmaceutical equipment into the Australian market, stating the attack was a response to a US military strike in Iran; Australia was not directly targeted or impacted. On 10 December 2025, ASD's ACSC and international partners released an advisory assessing that pro-Russia hacktivist groups were targeting CI by exploiting insecure internet-facing virtual network computing connections to reach OT systems. While generally less sophisticated than advanced persistent threat actors, their activities have resulted in operational disruptions including physical damage, with targeted sectors including water and wastewater systems, food and agriculture, and energy.

4. Techniques and Tradecraft

AI is improving cyber defence while also accelerating the speed, scale and sophistication of cyber threats. As Australian organisations integrate AI and large language models into their environments, ASD's ACSC has observed new and exploitable security risks emerging.

Social Engineering

Social engineering remains a longstanding and evolving threat that enables actors to compromise accounts, devices, systems or sensitive information. Recent advances in AI have amplified its effectiveness: actors can use AI to write phishing emails free from grammar and spelling errors and to automate targeted spearphishing. Communications may appear to come from a relative, colleague or senior manager, or from an authoritative part of the target's organisation, weaponising empathy, urgency and trust to induce staff to circumvent regular processes. The report is explicit that “AI lowers the skill barrier and increases scale, but does not fundamentally change the nature of social engineering attacks.”

The Australian Securities and Investments Commission has observed a sharp rise in online scams using deepfake videos of celebrities and politicians, with criminals increasingly exploiting topical news issues. In June 2026, an Australian broadcaster publicly reported that scammers had generated a fake version of its website, with misleading articles and false images of prominent Australians, to advertise a fraudulent investment platform.

The report sets out the attack in four stages:

•       The bait — actors research an organisation and its employees, with AI rapidly analysing publicly available information to create tailored approaches

•       The hook — social connection, imposed urgency or a disarming tone, with AI generating convincing emails, messages, images and voice recordings

•       The attack — users surrender information, click a malicious link or open a malicious attachment

•       The long game — the actor may disappear, or maintain access to an account for extended periods to conduct further activity

A case study records an August 2025 incident in which an Australian bank reported two phone calls to its internal helpdesk. The callers impersonated a Melbourne-based employee, citing the employee's ID number and claiming they could not connect with colleagues via Microsoft Teams. Their questions focused on a specialised financial services team in Melbourne and Sydney and demonstrated insider knowledge, indicating prior reconnaissance, and they obtained names, job titles and corporate phone numbers for several staff. The bank noted that the targeted team and individuals were not public figures, and that this lack of visibility made it easier for the callers to manipulate helpdesk staff, who did not immediately perceive the requests as suspicious.

Exploiting Vulnerabilities

AI may enable faster identification and analysis of weaknesses in systems and software, reducing the time, effort and skill required of malicious actors, and can accelerate the exploitation of vulnerability chains in which multiple weaknesses are linked to gain broader access. ASD's ACSC has previously observed that most vulnerability exploitation occurs within two weeks, with some CVEs successfully exploited within 48 hours of a patch or mitigation advice being released, and expects AI-driven tools to compress this window further over the next 12 months. “In the past, defenders had months to patch,” the report states, “now they have days, sometimes only hours.”

At the other end of the problem, ASD's ACSC recorded the successful exploitation this year of a vulnerability identified in 2008, demonstrating the need to patch legacy vulnerabilities as well as new ones.

Cybercriminals have been observed rapidly acquiring public and stolen data to extract usernames, passwords and other useful information, with some AI tools able to scrape social media and public sources to profile victims for identity theft, fraud and social engineering. The report cites public reporting that in 2025 a United States-based AI company disrupted a cybercriminal group using its AI to generate code for the theft and extortion of personal data at scale, semi-automating aspects of the operation from reconnaissance to network compromise, and analysing victims' financial data to determine ransom amounts.

Incidents mapped against the MITRE ATT&CK framework showed a notable shift away from reconnaissance techniques towards direct access and execution. Phishing remained the most observed initial access technique at 39%, followed by compromise accounts at 21%, user execution at 17% and valid accounts at 15%. Gathering victim identity information fell from 30% in FY2024–25 to 13% in FY2025–26. For CI-related incidents, the leading initial access techniques were phishing at 36%, valid accounts at 16% and drive-by compromise at 10%.

Compromising Supply Chains

The cyber security risks of suppliers and third parties are also an organisation's own risks, particularly where products or services hold privileged access or control a large portion of a supply chain.

In August 2025, numerous Australian organisations reported ransomware incidents to ASD's ACSC. Analysis of the combined technical data identified a known SonicWall vulnerability being actively exploited to deploy ransomware, with the activity matched to the Akira ransomware group. ASD's ACSC conducted scans identifying almost 1,000 potentially vulnerable devices and around 40 potentially vulnerable Australian organisations, many of them CI entities, then notified those organisations and published a public alert carrying vendor remediation guidance. Following publication, ASD's ACSC observed a reduction in reported incidents using similar techniques.

Since September 2025, ASD's ACSC has tracked the targeting of online code repositories, with actors gaining access through social engineering, compromised credentials and authentication tokens. They have been observed abusing legitimate tooling to modify public software packages and initiate supply chain compromises, extracting and publicly leaking credentials, migrating private repositories to public ones, and scanning for cryptographic secrets, passwords and sensitive keys. Exposed code bases give actors a better understanding of internal processes and systems, increasing an organisation's attack surface.

Over the same period ASD's ACSC has monitored a self-replicating malware known as Shai-Hulud, responsible for compromising hundreds of GitHub software packages. After compromising a repository maintainer's account, actors published a malicious version of a widely used package, which cascaded through downstream dependencies to produce rapid and widespread infection across the software supply chain. The malware harvests credentials and exfiltrates them to an actor-controlled endpoint, then creates a new public repository named Shai-Hulud under the victim's account and uploads the stolen secrets to it, publicly exposing them.

5. Named Operations and Enforcement

On 30 July 2025, ASD's ACSC and international partners released an updated joint advisory on Scattered Spider, which engages in data theft for extortion and has recently deployed ransomware alongside its usual TTPs. Its signature is layered social engineering conducted across multiple phone calls, culminating in spearphishing calls that convince IT helpdesk personnel to reset passwords or transfer multi-factor authentication tokens.

In March 2026, ASD's ACSC, the Kingdom of Tonga's National Computer Emergency Response Team and New Zealand's National Cyber Security Centre released a joint advisory on INC Ransom, a financially motivated group running a RaaS operation whose affiliates use double extortion — stealing data, encrypting files, and threatening publication via a data-leak site. Having previously targeted the United States and United Kingdom, the group has since early 2025 been increasingly observed targeting Australia, New Zealand and Pacific Island states. In mid-June 2025 the Tongan Ministry of Health was hit by a ransomware attack that rendered core services inaccessible and disrupted the national health care network; INC Ransom later claimed responsibility.

In November 2025, the Australian Government imposed targeted financial sanctions under the Autonomous Sanctions Regulations 2011 against two Russian cybercrime service providers, Media Land LLC and ML. Cloud LLC, and two of their key personnel, Aleksandr Alexandrovich Volosovik and Kirill Andreevich Zatolokin. The sanctions responded to the provision of ransomware infrastructure used to conduct attacks against Australian financial institutions, Australian businesses and their customers, and international CI, and were imposed in parallel with the United States and United Kingdom. This was the fifth activation of Australia's autonomous cyber sanctions framework, with penalties of up to 10 years' imprisonment for dealing with the sanctioned entities or individuals.

Conclusion

The Annual Cyber Threat Report 2025–26 presents a year in which the volume of cyber activity against Australia rose sharply while the character of the threat changed less than its speed. AI, the report states, has not introduced new cyber threats but has amplified and accelerated existing ones, compressing the window between disclosure and exploitation and lowering the skill and cost required to attack at scale. State-sponsored actors continue to pre-position within networks for use at a time of their choosing, and the distinction between their methods and those of cybercriminals continues to blur.

The full report is available from ASD's ACSC at cyber.gov.au.