AI-generated codebases show 92% vulnerability rate, researchers find

A recent security audit revealed that 92% of AI-generated codebases contain critical vulnerabilities, with an average of 8.3 exploitable findings per application. This alarming trend was highlighted by Microsoft's Security Response Center, which demonstrated how AI coding assistants like Claude Code can be manipulated through prompt injection to execute arbitrary shell commands, leading to security breaches. Similar vulnerabilities were found in IDEs like Cursor, where malicious code could be executed simply by cloning a repository. AI

IMPACT Highlights critical security risks in AI-assisted development, necessitating new security protocols and auditing practices for AI-generated code.

RANK_REASON Research paper detailing security vulnerabilities in AI-generated code. [lever_c_demoted from research: ic=1 ai=1.0]

Read on dev.to — Claude Code tag →

• Claude Code
• Cursor IDE
• CVE-2026-25592
• CVE-2026-26030
• DryRun Security
• Git
• Microsoft
• Security Response Center
• Semantic Kernel
• Sherlock Forensics
• Windows Calculator

• safety
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →