Skip to main content

Cybersecurity statistics / Governance

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

PublisherCloud Security Alliance (CSA) & Token Security
ReportAutonomous but Not Controlled AI Agent Incidents Now Common in Enterprises
Published21 April 2026
TopicsGovernance, Human Authorization, AI Agents, Enterprise

Published by Cloud Security Alliance (CSA) & Token Security in Autonomous but Not Controlled AI Agent Incidents Now Common in Enterprises , 21 April 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

80% of cybersecurity professionals say knowing when to override AI decisions is very important.
ISC2, 18/07/2026
82% of cybersecurity professionals say it is very important to establish when to trust AI outputs.
ISC2, 18/07/2026
95% of CISOs feel pressure to suppress or delay compliance-related security issues when business deadlines are at stake.
Checkmarx, 15/06/2026
0% of organizations ban vibe coding outright.
New Relic, 15/06/2026
88% of organizations include vibe coding in formal production policies.
New Relic, 15/06/2026
65% of CEOs believe they have a full AI inventory.
Veeam, 06/06/2026

Get the newsletter

Weekly cybersecurity statistics by email.