ServiceNow Report: 90% of Orgs Hit by Agentic AI Security Incidents

A new report from ServiceNow reveals that nine out of ten global organizations have suffered an agentic AI security incident within the last year. The study highlights a significant gap between perceived security readiness and actual defensive capabilities in the rapidly expanding field of autonomous AI systems.

While 36% of affected businesses contained these incidents in under 24 hours, the majority of organizations lack robust governance structures. According to the ServiceNow analysis, only 13% of companies maintain defensible security programs when measured against five critical domains: attack surface, accountability, security by design, visibility, and audit readiness.

Governance Gaps and Legacy Frameworks

Most agentic AI deployments are currently managed using legacy frameworks and human-driven systems. The ServiceNow report indicates that less than half of organizations utilize governance frameworks that cover key elements of AI agent deployments. This reliance on outdated methods leaves many businesses vulnerable to emerging threats associated with autonomous software agents.

Security teams often fail to integrate protection early in the development lifecycle. Just 44% of global security teams design security into an agent’s development before it is built. Consequently, the majority of organizations implement permissions and guardrails on an ad-hoc, after-the-fact basis, increasing the risk of misconfigurations and unauthorized access.

Scale and Operational Blind Spots

The scale of AI agent deployment is growing faster than organizational visibility. The ServiceNow data shows that 49% of organizations run between 100 and 999 AI agents, while 35% operate more than 1,000. This massive expansion creates significant blind spots for security teams trying to monitor and control autonomous behavior across complex IT environments.

Visibility remains a major challenge for many security operations. While 57% of security teams claim to have a full, accurate record of every AI agent in production and processes to track changes, 29% discovered agents running in production that had never been registered or known. These unregistered agents can operate as shadow AI, remaining unsupervised until they cause security issues.

What should you do?

Organizations must prioritize comprehensive inventory management and automated response capabilities to address the risks highlighted in the ServiceNow report. Security leaders should assess their current governance frameworks against the five key domains identified in the study to identify critical gaps in their agentic AI security posture.

What to do and how to stay safe: ServiceNow

• Conduct a complete audit of all AI agents in production to identify unregistered or shadow AI instances.
• Implement governance frameworks that specifically address the unique risks of agentic AI deployments.
• Design security controls into the development lifecycle of new AI agents before they are built.
• Develop automated response mechanisms to detect and mitigate anomalous agent behavior in real time.

Step-by-step guide: Prompt Injection Warning Signs: How to Spot AI Manipulation

General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

What percentage of organizations have experienced an agentic AI security incident?

According to the ServiceNow report, nine in ten global organizations have experienced an agentic AI security incident in the past year.

How many organizations have defensible agentic AI security programs?

Only 13% of organizations have defensible programs when measured against five key domains, including attack surface and visibility.

What is the main challenge in responding to agentic AI incidents?

Two-thirds of organizations cannot act automatically when an agent behaves anomalously, requiring human intervention to stop the issue.