Splunk Report: AI Agent Adoption Outpaces Security Oversight

A report published by Cisco on October 20 reveals that organizations are deploying AI agents faster than they can establish necessary security controls. The 2026 Splunk State of Observability study warns this rapid adoption leaves businesses exposed to security threats and operational errors.

According to the survey of 2,050 ITOps and engineering professionals, 38% of respondents stated their organization does not have clearly defined ownership for monitoring AI activities. This lack of accountability creates a significant gap in oversight as agentic AI tools become more prevalent in enterprise ecosystems.

Fragmented Responsibility Hinders Monitoring

The confusion over who is responsible for AI deployment stems from how these tools cut across traditional IT department boundaries. Responsibilities span software engineering, infrastructure management, web development, and information security, leading to fragmented oversight.

Consequently, 42% of those surveyed identified the main challenge in observing AI as the fact that a different team is responsible for it. Mark Woods, field CTO and chief technical advisor for EMEA at Splunk, noted that when observability is everyone’s job, it effectively becomes nobody’s job.

Managing Agents at Scale

IT and security teams are struggling to manage the unintended consequences of swift AI tool rollouts. While 85% of respondents said visibility into AI hallucinations is important, only 58% reported their organization can effectively monitor for these potential errors.

Identifying when an AI agent acts upon an error is difficult for many organizations, which may lack means to prevent such actions. A malicious attacker could potentially exploit an AI agent to trick it into exposing sensitive information, a risk that appropriate monitoring tools could mitigate.

Jeetu Patel, president and CPO at Cisco, stated that managing agents at scale requires reimagining both security and observability. He emphasized that questions about whether agents are compromised or performing well are inseparable and moving at machine speed.

What should you do?

Organizations should review their current AI deployment strategies to ensure monitoring procedures are established before or during rollout. Security teams need to define clear ownership for AI observability to prevent gaps in accountability.

What to do and how to stay safe: Splunk

• Define clear ownership and accountability for monitoring AI agents within your organization to prevent fragmented responsibility.
• Implement observability tools that can detect AI hallucinations and anomalous agent behavior in real time.
• Review access controls for AI agents to ensure they cannot be tricked into exposing sensitive information.
• Establish cross-functional coordination between software engineering, infrastructure, and security teams to manage AI deployment risks.

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 lack clear ownership for monitoring AI?

According to the Splunk State of Observability report, 38% of ITOps and engineering respondents said their organization does not have clearly defined ownership for monitoring AI.

What is the main challenge in observing AI according to the survey?

The main challenge cited by 42% of respondents is that a different team is responsible for the AI, leading to fragmented oversight across IT departments.

Can most organizations effectively monitor for AI hallucinations?

No, only 58% of respondents said their organization can effectively monitor for potential AI hallucinations, even though 85% consider visibility into them important.