Artificial intelligence is reshaping cybersecurity faster than many organisations can adapt, creating a growing divide between perceived security maturity and actual operational control.
That is the warning from Gigamon’s latest 2026 Hybrid Cloud Security Survey, which found AI is now involved in 83 percent of security breaches globally, while visibility gaps across hybrid cloud environments continue to undermine incident response and governance efforts.
For Australian organisations, the issue is becoming increasingly important as operational resilience obligations tighten under APRA CPS 230, CPS 234, and the SOCI Act.
The survey found 53 percent of Australian organisations experienced a breach in the past year, while 91 percent of local security leaders are reassessing hybrid cloud risk because of AI-driven threats.
According to the report, many organisations have built extensive security stacks and governance frameworks, but still struggle to achieve complete visibility into data in motion.
“Security efficacy is being measured by what has been implemented rather than what can be proven,” the report states.
That challenge is particularly acute in AI-enabled environments.
Modern enterprise systems increasingly rely on distributed workloads, APIs, AI agents, and hybrid cloud architectures that continuously move data between systems and regions. The report argues that traditional monitoring approaches were not designed for this level of complexity.
As a result, security teams often detect anomalies without being able to trace the full path of an interaction or determine root cause quickly.
Nearly one-quarter of organisations surveyed said they could not determine the root cause of breaches, while only 30 percent believed they had the tools required to respond effectively.
The issue is compounded by the rise of AI-powered attacks.
Almost half of respondents reported struggling with an increase in AI-driven threats, while 43 percent cited a shortage of cloud security expertise.
At the same time, organisations are increasingly automating security operations using AI. Ninety-four percent reported using AI systems to initiate security functions without human interaction, particularly around alert triage and prioritisation.
However, the survey warns that automation without visibility can create a false sense of confidence.
Gigamon positions “deep observability” as the solution — combining network-derived telemetry, packets, flows, and application metadata with traditional MELT data to provide a unified view across hybrid cloud environments.
According to the report, more than 90 percent of organisations now believe deep observability is foundational to securing AI deployments.
The findings also reinforce growing concern around encrypted traffic and post-quantum risk.
Today, malware hidden inside encrypted traffic accounts for billions of attack events annually, yet 76 percent of respondents still view encrypted data as inherently secure.
Meanwhile, 87 percent are concerned about future “harvest now, decrypt later” attacks, where threat actors collect encrypted data today with the expectation quantum computing may eventually render current encryption vulnerable.
The report concludes that organisations must move beyond fragmented monitoring and toward evidence-based security validation.
“Instead of approximating risk, they can validate it,” the report states in its discussion of network-derived telemetry and deep observability.
