AI Agents Deployed Faster Than Security Teams Can Control Them

Security visibility lags behind AI agent deployment

Enterprises are rolling out autonomous AI agents at a pace that far exceeds their ability to secure them, creating what researchers describe as a "shadow workforce" of non-human identities operating beyond IT oversight.

Rubrik Zero Labs surveyed more than 1,600 IT and security leaders and found that 86% anticipate AI agents will surpass their organization's security guardrails within the next year. More troubling: only 23% of respondents claim full visibility into the AI agents currently operating in their environments, according to findings first reported by TahawulTech.

The gap between deployment speed and security readiness means organizations cannot adequately secure the identities that are actively making decisions and interacting with critical data. These autonomous systems are proliferating faster than enterprises can track them, creating new attack surfaces that traditional security controls were not designed to address.

Why it matters

This research quantifies a fundamental mismatch between AI adoption velocity and security maturity. When the vast majority of security leaders expect their controls to be outpaced within 12 months, it signals that current governance frameworks are inadequate for autonomous systems. The inability to roll back agent actions without system disruption means organizations are accepting operational risk they may not fully understand.

Efficiency gains remain elusive

The promise of AI agents reducing manual workload has not materialized for most organizations. More than 80% of survey respondents report that AI agents require more manual oversight than the efficiency they deliver. This finding challenges the core value proposition of agentic AI systems in enterprise settings.

Additionally, 88% of respondents lack the ability to roll back agent actions without disrupting systems. This limitation means that when an AI agent makes an error or is compromised, organizations face a choice between accepting the consequences or risking broader operational impact through remediation efforts.

Attack surface concerns mount

Nearly half of the surveyed security leaders expect agentic systems to drive the majority of cyberattacks in the coming year. This expectation reflects growing awareness that AI agents represent both a new category of asset to protect and a new vector for adversaries to exploit.

The research underscores that AI strategy cannot be separated from resilience strategy. Boards and executive teams must integrate security considerations into AI deployment decisions from the outset, rather than treating governance as an afterthought once systems are operational.

The findings point to an urgent need for enterprises to develop visibility tools, identity management frameworks, and rollback capabilities specifically designed for autonomous AI systems before the security gap widens further.

The research was conducted by Rubrik Zero Labs and details were first reported by TahawulTech.