Reco research finds 80% of AI tools run without IT oversight

Security vendor Reco has published research warning that the AI agent ecosystem is increasingly risky due to gaps in IT oversight, a surge in published vulnerabilities, and Model Context Protocol (MCP) security risks.

Analysing anonymised telemetry from large enterprises, public MCP servers and National Vulnerability Database disclosures, Reco found that 80% of AI tools operate with no oversight, and SMBs have an estimated 414 unsanctioned tools per 1,000 employees.

The report also examined 500 MCP servers, finding half can execute shell commands directly, more than eight in ten can read or write local files, and roughly three-quarters can make outbound network calls. Over a quarter expose a network endpoint, and half of those ship with no authentication.

Reco tracked 637 vulnerabilities across agents and LLM tools, with 525 disclosed in the past 18 months, including at least 111 rated critical with CVSS scores of 9.0 or higher. The average monthly disclosure rate rose from under five during 2023 and 2024 to around 29 since January 2025.