Splunk report: businesses are deploying AI agents faster than they can monitor them
Cisco-owned Splunk's annual 2026 State of Observability report found that 38% of the 2,050 IT and engineering professionals surveyed said their organization still lacks clearly defined ownership for monitoring AI, while 42% said their main challenge was that this responsibility is split across different teams, including software engineering, infrastructure, and security. While 85% said it was important to detect when an AI model hallucinates, meaning it confidently produces incorrect information, only 58% said they could actually monitor for it effectively, a gap that leaves room for an attacker to trick an agent into exposing sensitive information unnoticed. Splunk's EMEA field CTO Mark Woods summed it up by saying that "when agent observability is everyone's job, it becomes nobody's job." On the other hand, 86% of respondents said AI had helped them detect IT or security issues faster, though 14% said agents actually made problems harder to detect, mainly because of the extra time needed to double-check whether what an agent reports is accurate. For a business just starting to roll out AI agents internally, the report shows the risk isn't only technical but organizational: if no one is explicitly responsible for watching what an agent does, problems surface late.
Sources
This summary was written automatically with AI from the public sources listed above and checked by an independent AI reviewer before publication. For full details, read the original source.