68% of enterprises have zero technical visibility into autonomous AI agent workflows that inherit user identity and single sign-on tokens.

71% of organizations reported between one and nine intrusions, up from 47% the previous year.

2% of organizations reported more than 10 intrusions, unchanged from the previous year.

89% of organizations expect increased regulation within five years or less, up from 66% in 2025.

Approximately 23% of organizations have visibility into about half of their OT environment.

19% of adults aged 18+ in the United States, UK, Austria, Germany, and Switzerland have experienced AI-driven identity harm, rising to 30% among Gen Z.

72% of organizations prioritize machine identity at scale when evaluating identity infrastructure.

53% of victims with no financial loss reported a resolution.

Understaffing in European cybersecurity roles is 48%, which is 14 percentage points higher than in the rest of the world.

82% of organizations experienced at least one production failure tied to AI-generated code in the past six months.

5% of organizations restrict vibe coding to non-production environments.

61% of leaders rate AI-generated code as somewhat higher quality, 33% rate it as much higher, and 2% perceive it as lower quality.

24% of organizations reported intrusions in both IT and OT systems, down from 60% in 2025 and the lowest since 2022.

13% of adults aged 18+ in the United States, UK, Austria, Germany, and Switzerland have created a family codeword to guard against likeness theft.

52,185 threats were hosted on domains that enterprise security stacks are configured to trust, including Google Drive, Dropbox, and SharePoint.

96% of developers acknowledge having AI tooling integrated in their IDEs.

74% of organizations lack a unified view of sensitive data and the identities that can access it.

Account-problems scams are the highest-volume scam type, with 74% of victims sharing high-value personally identifiable information (PII).

78% of enterprise servers are reachable over SMB or WinRM, administrative protocols commonly exploited for ransomware spread and lateral movement.

Manus AI accounted for 15.6% of attributed AI site builders, Blackbox AI accounted for 14.3%, and Anything AI accounted for 9.8% of attributed builders.