Australian organisations are moving to deploy AI agents faster than they can secure them, with 88 per cent expecting autonomous systems to outpace their security safeguards within the next 12 months, according to new research from Rubrik Zero Labs.
The findings, drawn from a survey of more than 1,600 IT and security leaders, point to governance and recovery gaps emerging as businesses operationalise agentic AI. Only 22 per cent of respondents reported having full visibility into the agents operating in their environments, a level the report suggests may be overestimated.
The report argues that limited visibility can translate into weak identity governance, as AI agents and other automation tools rely on non-human identities that can accumulate access privileges. Rubrik Zero Labs described this identity growth as a “shadow workforce”, warning that persistent access and limited oversight could create pathways for misuse, compromise and lateral movement.
Operationally, the study suggests many organisations are not yet seeing net efficiency gains. It found 80 per cent of respondents said AI agents require more manual oversight than the efficiency they deliver, while all Australian respondents said they lack the ability to roll back agent actions without system disruption.
The report also points to rising concerns about cyber resilience as agentic systems become more widespread. It said 96 per cent of leaders expressed concern about meeting recovery objectives as agent-driven threats increase.
On the threat outlook, 42 per cent of Australian respondents expected agentic systems to drive the majority of attacks in the coming year, reflecting what the report characterised as a shift in attacker capabilities as automation compresses timelines and scales activity.
“AI adoption is outpacing our ability to control it. Enterprises are struggling because they’ve deployed systems they can’t fully observe, govern, or restore,” said Kavitha Mariappan, chief transformation officer at Rubrik. “We have to move past the debate of whether AI is risky and address the harder reality: as decision-making shifts from human to machine, the critical challenge for every leader is maintaining operational safety in an increasingly autonomous landscape.”
The research, titled The State of the Agent: Understanding Adoption, Risk, and Mitigation, combines survey results with technical analysis of attack vectors across what it describes as tool, cognitive and identity layers of AI systems.
