AI drives sharper cyber threats, tech leaders warn

Technology executives have warned that organisations face a sharper cyber threat as artificial intelligence reshapes attack methods. Their comments come during Cyber Security Action Month.

The remarks reflect a common concern across the sector. Companies are deploying more AI tools, while attackers are using AI to increase speed, scale and automation.

Security leaders said the issue now extends beyond awareness campaigns and staff training. Organisations need stronger governance over AI systems, better visibility across data and infrastructure, and regular testing of incident response.

Mark Buckley, Vice President, Australia & New Zealand, Genesys, focused on the controls needed for increasingly autonomous AI agents inside organisations.

"As agentic AI becomes more autonomous, security, privacy and governance are what underpin customer trust. AI agents can take action across enterprise systems, making it critical for organisations to define what AI can do, what it can access and when people need to step in. Governance can't be an afterthought; it needs to be built in from the start. As these capabilities advance, agentic orchestration can help organisations coordinate AI, people, systems and data while maintaining the visibility and control needed to do so responsibly. Leaders need to understand where risks are emerging, technology teams need the right permissions and guardrails, and employees need to know how to use AI responsibly and where human judgment matters. As AI evolves, organisations will need to continually reassess whether their controls are keeping pace. Cybersecurity Awareness Month is a timely reminder that responsible AI depends on getting this foundation right. Strong security and governance can give organisations greater confidence to innovate while protecting customer trust in every interaction," said Mark Buckley, Vice President, Australia & New Zealand, Genesys.

His comments reflect a broader shift in cyber security discussions. The debate now covers not only external threats, but also the internal permissions, access boundaries and oversight structures tied to AI systems that can make decisions or trigger actions.

Data Signals

Greg Taylor, SVP & GM, Confluent, argued that many security tools will fall short if organisations lack timely, connected data.

"Cyber Security Action Month is a reminder that attackers are not waiting for organisations to catch up. They are already using AI to move faster and find weaknesses at an unprecedented scale. Businesses are implementing AI-powered threat detection tools to help security teams respond faster, but many lack the data foundations needed for those tools to work effectively. To detect abnormal activity, organisations need a robust baseline of normal activity and a continuous, live stream of signals that inform the system of updates, access requests and data transfers. Tools then use behavioural analytics to detect suspicious activity and alert the relevant teams. When data is delayed or fragmented, malicious activity won't be surfaced until after the damage has been done. This is why real-time data streaming is important. It gives security teams and their tools the live context needed to connect activity across systems and determine whether a change is routine or a sign of something more serious. As attackers continue to move faster, organisations cannot afford to make security decisions on yesterday's data. Leaders need to treat the data layer as part of their security architecture, because the faster teams can connect the signals, the faster they can act before an anomaly becomes a breach," said Greg Taylor, SVP & GM, Confluent.

This highlights a practical problem for many businesses. Security systems often pull information from separate platforms, and delays in that flow can leave teams reacting after an incident has spread.

Preparedness Test

Charles Spencer, General Manager ANZ, Vectra, said the threat environment is already severe and is becoming more complex as AI models improve.

"Cybersecurity Awareness Month feels particularly timely this year. Recent developments involving AI agents have raised fresh questions about the security controls needed to manage these technologies, particularly in sensitive environments. The question is no longer whether AI will change the threat landscape, but whether organisations are prepared to detect and contain what comes next. Ensign's 2026 Cyber Threat Landscape Report found that frontier AI models can already complete multiple stages of a simulated cyber attack. Meanwhile, the capability gap between the leading Eastern and Western models tested is narrowing, with a leading Eastern model delivering comparable offensive performance at approximately one-fifth the operating cost. These findings come against an already serious threat environment. Ransomware activity across Australasia increased by more than 600%, while the sale of initial access rose fivefold. Organised crime accounted for 52% of observed threat activity. Awareness remains important, but it cannot be treated as an annual exercise. At Vectra, we encourage Australian organisations to move beyond awareness to preparedness. Businesses need to assume they will be probed and potentially compromised, then continuously test whether they can detect, contain and recover from an AI-enabled attack," said Charles Spencer, General Manager ANZ, Vectra.

Taken together, the comments show how AI is changing both sides of cyber security. Companies are adopting AI across operations and customer systems, while threat actors are using similar advances to carry out faster, cheaper attacks. That combination is pushing security teams to focus on governance, live data and regular resilience testing.