47% of unknown AI agents emerge in LLM platforms, including custom tools, assistants, and plugins.

55.0% of consumers are not comfortable with AI agents making purchases on their behalf

88% of IT and security leaders say they lack the ability to roll back AI agent actions without system disruption.

15% of enterprises report that 76–100% of AI agents have defined ownership.

34% of enterprises report ownership visibility for just 26–50% of their AI agents.

54% of enterprises report having between 1 and 100 unsanctioned AI agents.

47% of enterprises experienced a security incident involving an AI agent in the past year.

Only 21% of enterprises have formal decommissioning processes for AI agents.

40% of unknown AI agents emerge in SaaS tools with built-in automation.

13% of enterprises report fully autonomous AI agent models.

38% of enterprises require human approval when AI agents exceed their scope.

24% of enterprises require actions that exceed an agent's scope to be logged.

53% of enterprises use human authorization as a primary signal for governing AI agent behavior.

Over 80% of security teams at enterprises lack full visibility into the applications and AI agents created by business users.

86% of IT and security leaders expect AI agents to outpace their organization’s security guardrails within the next year.

Nearly half of IT and security leaders expect agentic systems to drive the majority of attacks in the coming year.

Nearly nine in ten IT and security leaders express concern about meeting recovery objectives as agent-driven threats increase.

68% of enterprises report high confidence in their visibility into AI agents.

82% of enterprises have discovered previously unknown AI agents in the past year.

40% of unknown AI agents emerge in developer-created workflows.