In 2025, an AI agent placed in the top 5% of teams in a major cybersecurity competition.

82% of malicious files have unique hashes that traditional pattern-matching fails to detect.

14.4% of AI agent configuration files grant arbitrary code execution permissions for Node.js.

Organizations delivering expertise through open collaboration channels increased by 29%.

The Internet Watch Foundation detected 26,362% more AI videos of CSAM in 2025 compared to 2024.

Less than one-fifth of total healthcare email incidents involved identity abuse via stolen credentials, yet these remained the most damaging type of attack.

Live retail fraud dropped by 69% month-over-month in November 2025 while non-live fraud rises.

630,000 healthcare patient records were exposed due to stolen login credentials alone.

44% of organizations use or plan to use static API keys and 43% use or plan to use username/password combinations for agents.

21% of organizations maintain a real-time registry or inventory of their agents.

Almost 20% of developers let AI automatically save changes to the project's main code repository without human review.

84% of organizations doubt they can pass a compliance audit focused on agent behavior or access controls.

Over 70% of organizations expect to manage dozens to hundreds of agents within the next 12 months, with 39% expecting 1–100 agents and 31% expecting 101–500 agents.

24.9% of Gemini usage occurs through personal accounts.

The top 1% of early adopter organizations use more than 300 GenAI tools.

Across the top 100 most-used GenAI SaaS applications, 82% are classified as medium, high, or critical risk.

In companies leading in AI adoption, nearly 90% of developers use AI coding assistants.

32.3% of ChatGPT usage occurs through personal accounts.

Developers at frontier companies are 11.5× more likely to use AI coding assistants than developers in low-adoption environments.

Cautious enterprises typically employ fewer than 15 GenAI tools.