AI Code Assistants Leak Secrets at Double the Human Rate, Report Shows

AI-assisted commits are exposing sensitive credentials at roughly twice the rate of human-written code, according to GitGuardian's State of Secrets Sprawl 2026 report. The analysis found that code commits co-authored by AI tools like Claude Code leaked secrets at 3.2% compared to 1.5% for baseline human-written commits—a doubling that reflects how AI acceleration outpaces security governance in development workflows.