Australia says expect more outages as hackers use AI to break in

Australian internet users should prepare for more frequent disruptions to essential online services. The Australian Signals Directorate (ASD) issued a warning on Monday that system security must take precedence over system availability. This means websites, banking apps, and government portals may go offline without notice while operators patch vulnerabilities.

The agency released its annual cyber threat report on Tuesday. The data shows a sharp rise in digital attacks over the past year. ASD responded to 1,500 cybersecurity incidents during the 2025-26 period. That figure is a 20% increase from the previous year. The agency also notified businesses, governments, and critical infrastructure operators about more than 4,400 potential malicious incidents. That number represents an 80% rise in one year.

Abigail Bradshaw, the ASD director general, stated that hostile foreign governments, criminals, and hackers are using artificial intelligence to gather intelligence and disrupt infrastructure. She explained that these actors use AI to steal money and find weaknesses in systems faster than before. Bradshaw said on Monday that Australia needs AI tools to automate defenses and stay ahead of attackers. She added that the country must accelerate the modernization of its technology foundations to protect daily systems.

The report notes that state-based actors, particularly in the Indo-Pacific, launched persistent breach attempts against Australian networks last year. ASD claims these actors are preparing for future conflicts. The agency describes these actions as future intelligence collection or disruption activities. It warns that all network owners are facing elevated threat levels. Cybersecurity is now a core business risk and a leadership responsibility, not just a technical issue.

Government officials emphasized the need to balance risk with progress. Richard Marles, the defence minister, said Australians must enhance cyber defences while using AI to their advantage. He made these comments when the report was released. Tony Burke, the minister for home affairs and cybersecurity, said the government is positioning Australia to respond to emerging threats and opportunities. Both ministers noted that collaboration with leading AI labs is critical.

ASD identified legacy technology as a major risk factor. Outdated systems and devices that no longer receive updates from creators are attractive targets for hackers. The agency urged organizations to update their systems and report suspicious incidents. It also advised companies to seek help from ASD when needed. Bradshaw and Marles reiterated that keeping a system online is less important than securing it. This approach will likely result in more planned outages to install fixes.

Last week, new vulnerabilities in PHP, a programming language used by most websites, forced a rapid response from governments and businesses. The incident demonstrated how quickly gaps can appear in widely used code. ASD said integrating AI into cybersecurity functions can help find these weaknesses and suspicious behavior. The agency believes that using the same tools that attackers use is the best way to protect against them.

Bradshaw mentioned that ASD has access to Anthropic’s Mythos model. This frontier AI tool can identify unknown flaws in IT systems. The lab said it could not release the model publicly because of its power. Australian government agencies believe having access to such models is necessary to bolster defenses. Bradshaw concluded that in a threat environment changed by AI, collective action is the strongest defense available to the nation.