45% of AI-generated code samples failed security tests, frequently introducing OWASP Top 10 vulnerabilities due to a lack of security-by-design principles.

Nearly 22% of code generated by open-source models contained package names that do not exist, creating a direct vector for supply chain attacks.

1 in 5 organizations inadvertently expose internal applications or secrets. A common flaw is authentication logic living entirely in the client-side browser.

Includes poor dependency hygiene, lack of code review, and syntax errors. 48% of generated snippets overall contained some form of vulnerability.