Cisco Highlights Concerns Over Risks Posed by Autonomous AI Agents

A 2026 State of Observability report found that 83% of respondents are concerned about the unintended consequences of AI agents. The report noted that while 87% of technology professionals prioritize AI monitoring, 56% cannot measure the quality of responses generated by large language models. Cisco highlighted these difficulties, stating that agents can execute software commands and access company information.

The findings indicate growing pressure on engineering, operations, and cybersecurity teams to ensure AI agents comply with organizational safeguards. Because agents can make decisions and trigger operational actions, monitoring efforts are crucial to prevent inaccurate outputs from becoming critical failures.