Skip to main content

Cybersecurity statistics / AI Agents

90% of respondents agree that AI agents should operate under least privilege principles, with bounded access and tighter controls.

PublisherPalo Alto Networks
Report2026 Identity Security Landscape
Published14 May 2026
TopicsAI Agents, Least Privilege

Published by Palo Alto Networks in 2026 Identity Security Landscape, 14 May 2026. The figure is taken from the report as published; the full methodology is in the source.

View the original report

Compare this across sources

12 sources answer what share of organisations have an ai governance policy in place. They report between 18% and 44%, with a median of 36.5%.

See all 12 sources

Related statistics

The typical enterprise environment contains 1.17 AI agents per device.
Vectra AI, 25/07/2026
The most extreme enterprise environment contains 96 AI agents assigned to a single device.
Vectra AI, 25/07/2026
35% of enterprise environments contain more AI agents than devices.
Vectra AI, 25/07/2026
Only 21% of organizations have implemented controls for non-human identities (NHIs).
JumpCloud Inc., 18/07/2026
Non-human identities outnumber human users in 83% of organizations.
JumpCloud Inc., 18/07/2026
The share of organizations requiring human review before high-risk AI actions dropped from 40% to 25% in six months.
JumpCloud Inc., 18/07/2026

Get the newsletter

Weekly cybersecurity statistics by email.