Cisco Splunk 2026 State of Observability: AI agent rollouts outpace monitoring and security oversight
The Cisco Splunk 2026 State of Observability report, published on October 20, finds that organizations are deploying AI agents faster than they can build monitoring procedures and security controls. This leaves them exposed to security threats, hallucinations and related problems. Of the 2050 ITOps and engineering respondents, 38% said their organization has no clearly defined ownership and accountability for monitoring AI. Responsibility is spread across software engineering, infrastructure, web development and information security. The report is survey-based and describes no specific incident, vulnerability or threat actor.
Detalhes técnicos
- 42% of respondents said their main challenge in observing AI is that a different team is responsible for it.
- 85% said visibility into AI hallucinations is important, but only 58% said they can effectively monitor for them.
- Teams often cannot identify when an agent is acting on an error, and may have no means of preventing the action.
- The article gives a hypothetical risk: an attacker could trick an AI agent into exposing sensitive information.
- 86% said AI deployment improved the speed of detecting IT or security issues, and 36% called the improvement "significant". 14% said agents made issues harder to detect and resolve, mainly because of the extra time spent double-checking agent output.
Remediação
- Assign clear ownership and accountability for AI agent monitoring.
- Converge observability and security data on common data platforms to detect and respond to agent issues in real time, as Cisco and Splunk executives recommend.
- Strengthen collaboration between security and IT operations teams.
Linha do tempo
-
Cisco Splunk 2026 State of Observability: AI agent rollouts outpace monitoring and security oversight
ler a análise desta fonte ocultar
The Cisco Splunk 2026 State of Observability report, published on October 20, finds that organizations are deploying AI agents faster than they can build monitoring procedures and security controls. This leaves them exposed to security threats, hallucinations and related problems. Of the 2050 ITOps and engineering respondents, 38% said their organization has no clearly defined ownership and accountability for monitoring AI. Responsibility is spread across software engineering, infrastructure, web development and information security. The report is survey-based and describes no specific incident, vulnerability or threat actor.
Technical Details
- 42% of respondents said their main challenge in observing AI is that a different team is responsible for it.
- 85% said visibility into AI hallucinations is important, but only 58% said they can effectively monitor for them.
- Teams often cannot identify when an agent is acting on an error, and may have no means of preventing the action.
- The article gives a hypothetical risk: an attacker could trick an AI agent into exposing sensitive information.
- 86% said AI deployment improved the speed of detecting IT or security issues, and 36% called the improvement "significant". 14% said agents made issues harder to detect and resolve, mainly because of the extra time spent double-checking agent output.
Remediation
- Assign clear ownership and accountability for AI agent monitoring.
- Converge observability and security data on common data platforms to detect and respond to agent issues in real time, as Cisco and Splunk executives recommend.
- Strengthen collaboration between security and IT operations teams.