What is the main challenge in monitoring AI agents?

Only 23% globally (38% in Japan) reported fully monitoring active AI agents in their IT environments, indicating a significant visibility gap.

What is the concept of 'Shadow Workforce'?

It refers to the rapidly increasing 'non-human identities' that companies cannot keep up with in terms of tracking and governance, posing new attack vector risks.

Does the implementation of AI agents contribute to operational efficiency?

Over 80% of respondents (86% in Japan) stated that the management burden by human hands is greater than the efficiency gained from implementing AI agents.

What is the recovery capability for AI agent malfunctions?

88% of organizations (84% in Japan) lack the ability to roll back specific actions of AI agents without resetting the entire system.

How is the future of cyber attacks expected to change?

About half of the respondents expect that within the next year, most attacks will be conducted using agent-based systems, leading to faster and larger-scale attacks.

Rubrik Warns of Growing Security Gaps as AI Agent Adoption Accelerates

Published May 19, 2026 8:10 PM ・ Updated Jun 14, 2026 7:42 AM ・ 5 min read ・ Source: PR TIMES

A Rubrik Zero Labs report indicates that AI agent adoption is outpacing security measures. 86% of IT leaders expect AI agents to overwhelm security guardrails within a year, while visibility gaps and the rise of a 'shadow workforce' create significant risks and recovery concerns.

Rubrik, a leader in security and AI operations, today released the latest research report from Rubrik Zero Labs. The study reveals that AI agent adoption in enterprises is moving faster than the development of security measures. Organizations are operating agent-based systems without sufficient governance controls, leading to a gap between innovation and security.

The survey of 1,600 IT and security leaders revealed key findings: 86% of respondents (83% in Japan) predict that the surge in AI agents will surpass their security guardrails within the next year. Only 23% (38% in Japan) reported being able to fully monitor active AI agents in their environment. As a result, identities linked to critical data and decision-making are not being adequately protected.

This challenge is worsened by the explosion of identities. Non-human identities associated with agents are increasing faster than corporate tracking can keep up, a phenomenon researchers call the 'shadow workforce.' These identities often operate with persistent access and limited oversight, creating new pathways for misuse, breaches, and lateral movement.

Expectations for AI agent operations are also wavering. Over 80% (86% in Japan) stated that the manual management burden outweighs the efficiency gained from agent adoption. Furthermore, 88% (84% in Japan) reported lacking the capability to undo specific actions taken by AI agents without resetting the entire system. Approximately 90% expressed concern about meeting recovery goals.

The threat landscape is accelerating, with half of respondents expecting agent-based systems to account for the majority of attacks within a year. AI agents are shortening attack times and increasing scale, blurring the lines between internal risks and external breaches.

Kavitha Mariappan, Chief Transformation Officer at Rubrik, stated: 'AI adoption is outpacing the ability to control it. As decision-making shifts from humans to machines, maintaining operational safety in autonomous environments is a critical challenge.' AI strategy is now inseparable from resilience strategy, making identity verification and visibility prerequisites for safe agent system adoption.

FACT BOX

Source: PR TIMES
Category: Survey
Organizations: Renown Health

Rubrik Warns of Growing Security Gaps as AI Agent Adoption Accelerates

⚡ Key Points

FACT BOX

Editorial & Verification Standards

FAQ

Cite this article — HOW TO CITE

AI CRAWLER ACTIVITY