AI writes half our code now. It still fails security tests 44% of the time.
SMRTR summary
Veracode's 2026 GenAI Code Security Report reveals that AI models now write compilable code nearly 100% of the time, but still fail security checks 44% of the time — a rate unchanged over a year of testing. With AI now authoring roughly half of all committed code, that static failure rate represents a massively growing attack surface. Reasoning models score slightly better, but even the top model, GPT-5.5, fails one in three security tasks.
SMRTR provides this summary for quick context. The original article belongs to The Next Web.
Read the original article