8% of enterprises say AI agents never exceed their intended permissions.

61% of security professionals want a "strategic pause" to calibrate defenses against AI-driven threats, up from 48% last year.

A year ago, 55% of AI models failed basic vulnerability research and 93% failed exploit development tasks

18.39% of organizations have exposed OpenAI credentials.

11.92% of organizations have exposed Databricks credentials.

51% of organizations in Canada identify AI model monitoring, auditing and assurance tools as a priority.

49% of enterprise leaders anticipate a material AI-agent–driven security or fraud incident within six months.

97% of enterprise leaders expect a material AI-agent–driven security or fraud incident within 12 months.

41.88% of production organizations have leaked AI or ML credentials.

47% of organizations view AI as a top future threat to unstructured data, and 40% view AI as a core security capability for unstructured data.

28.49% of organizations have exposed Hugging Face tokens.

53% of cybersecurity professionals identify lack of visibility into AI, automation and machine access as their top risk.

10.10% of organizations have exposed Anthropic credentials.

1% of professional pentesters believe AI-only scanning is effective for uncovering high-impact, exploitable vulnerabilities.

31% of organizations do not know whether they experienced an AI security breach in the past 12 months.

Autonomous agents account for more than 1 in 8 reported AI breaches.

More than 40% of organizations allocate less than 10% of their budgets to AI security.

91% of organizations added AI security budgets for 2025.

AI benefits attackers over defenders by a 13-to-1 ratio, with 78% saying AI makes attacks more effective versus 6% saying AI has improved defenses.

78% of security leaders say AI has made ransomware attacks more effective.