Australian Signals Directorate urges businesses to adopt AI defence against faster attacks

Australia's cyber intelligence agency has released its Annual Cyber Threat Report, warning that malicious groups and hostile actors are using artificial intelligence to hunt for system weaknesses, steal data and disrupt operations at greater speed and scale than ever before.

Abigail Bradshaw, the director-general of the Australian Signals Directorate, said on Monday that organisations now need automated AI tools within their own defences to keep pace with AI-enabled attackers. She also noted that businesses must prepare for regular service interruptions, arguing that companies should prioritise fixing security issues over keeping digital systems continuously online.

Outages and automated attacks on the rise

The report shows the agency responded to 1,500 cybersecurity incidents across Australia over the 2025–26 financial year, an increase of 20 per cent. It sent more than 4,400 warning notifications about potential malicious activity to outside organisations, representing an 80 per cent rise in a single year. The agency received a cybercrime report every six minutes.

According to the directorate, unsupported legacy software remains the single largest technical vulnerability across Australian workplaces. Criminals and state-backed operators are using advanced machine learning models to scan public-facing systems for known vulnerabilities in minutes, cutting down the window of time IT teams have to apply security patches.

The report also pointed to operational risks from businesses giving AI agents wider access to corporate data and networks, where unintended automated actions can create new vulnerabilities.

What this means for Australian businesses

Small and medium business owners do not need huge cybersecurity budgets to respond to these warnings, but standard operating practices need updating.

• Audit and retire unsupported software, because attackers use automated AI scanners to target known flaws in outdated tools first.
• Enable automatic security updates and patch critical software immediately, even if taking an internal system offline briefly causes a short delay.
• Turn on AI-assisted security features included in existing business software, such as automated phishing detection and unusual login flagging in email suites.
• Check the permissions granted to AI tools and workplace agents so they cannot access financial systems or export sensitive customer databases without approval.
• Enforce multi-factor authentication across all staff logins to stop automated credential stuffing and account takeover attacks.