AI-generated code security rose from 56% to 62% in Veracode’s latest analysis, while the proportion of code containing vulnerabilities fell from 44% to 38%. The six-percentage-point improvement is the largest since the company began its GenAI Code Security Report series in July 2025. Anthropic’s Claude Opus 5.5 and OpenAI’s GPT-6-Sol topped the rankings, each achieving a 73% security score.
Veracode assessed 18 models across Python, C#, JavaScript and Java, testing four Common Weakness Enumeration (CWE) categories: CWE-89, CWE-327, CWE-80 and CWE-117. Ten models scored above 60%, compared with four in the previous report. However, the lowest results remained between 44% and 50%, indicating that vulnerability levels still vary according to model choice.
“We’ve been analysing code generated by multiple LLMs for more than a year. Until now, we have not seen a marked improvement in AI code security, despite consistently near-perfect functionality,” said Chris Wysopal, Co-founder & Chief Security Evangelist at Veracode. “This is the first time since running our study that we’ve been able to report a meaningful move on the security dial. Ten models now clear 60 percent, up from four in the last report, and the overall average has jumped six percentage points. With around half of code now generated by AI, today’s results are truly encouraging.”
Cross-site scripting drives security gains
Claude Opus 5.5 and GPT-6-Sol’s joint score of 73% surpassed the previous high of 68%
Claude Opus 5.5 and GPT-6-Sol’s joint score of 73% surpassed the previous high of 68%, recorded by GPT-5.5 in July 2026. Anthropic and OpenAI each had models among the top seven. In July, six of 11 models scored between 50% and 53%; in the latest results, only two remained in that range.
Cross-site scripting (XSS), represented by CWE-80, was the main contributor to the overall improvement. Its score rose to 18%, the biggest change in the dataset. Crypto vulnerabilities (CWE-327) remained at 87%, SQL injection (CWE-89) at 84%, and log injection (CWE-117) at 13%.
Veracode said XSS can be addressed through pattern recognition, while log injection often requires data-flow reasoning across multiple functions. The latter remains a challenge for language models.
Model costs and language differences
The highest-scoring models also carry premium compute costs. Large models averaged 55% on security tasks, compared with 51% for small models, reversing last year’s finding that size had “essentially no effect” on security. General-purpose models averaged 54%, ahead of coding-specialised models at 51%; the gap has widened from one percentage point in July.
“The models that produce the most secure code are the ones that cost the most to run at scale, yet they still fail on more than one in four security tasks,” Wysopal said. “The answer should never be to use the most expensive model and hope for the best. A mid-tier model at a 60 percent pass rate plus Veracode scanning is more accurate, more cost-effective, and more scalable than a frontier model at 73 percent with no verification. Organisations must look for a combination of coding model, security testing, and remediation to get the highest security outcome.”
Java remained an outlier, with a 31% score, just above its 30% result in July. Python scored 64%, C# 60% and JavaScript 57%, leaving a 26-to-33-percentage-point gap between Java and the other languages. Overall, around four in 10 AI code tasks still produced vulnerable code without security-specific guidance.
Security verification remains necessary
Veracode’s 2026 State of Software Security report found that 82% of organisations have security debt, defined as vulnerabilities unresolved for more than a year. It also reported that 66% of the most dangerous long-lived vulnerabilities originate in third-party code and take an average of 358 days to fix.
“Newer LLMs are producing more code that’s both functional and secure, but that does not mean we can start blindly trusting AI-generated code,” Wysopal closed. “The most effective strategy isn't chasing the most expensive model for marginal gains; it's combining AI code generation with continuous, independent security verification that catches what the models miss. That approach covers the full risk surface, from first-party code to third-party dependencies.”
Veracode, the pioneer in application risk management, announces a jump from 56 percent to 62 percent in overall AI-generated code security, alongside a drop in vulnerabilities from 44 percent to 38 percent — the most significant improvement since the GenAI Code Security Report series began in July 2025. Anthropic's Claude Opus 5.5 and OpenAI's GPT-6-Sol co-lead the rankings at 73 percent, surpassing the previous high of 68 percent set by GPT-5.5 in July 2026.
“We’ve been analysing code generated by multiple LLMs for more than a year. Until now, we have not seen a marked improvement in AI code security, despite consistently near-perfect functionality,” said Chris Wysopal, Co-founder & Chief Security Evangelist at Veracode. “This is the first time since running our study that we’ve been able to report a meaningful move on the security dial. Ten models now clear 60 percent, up from four in the last report, and the overall average has jumped six percentage points. With around half of code now generated by AI, today’s results are truly encouraging.”
Meaning vulnerability volume
Veracode tested 18 models across four languages (Python, C#, JavaScript, Java) and four Common Weakness Enumeration (CWE) categories (CWE-89, CWE-327, CWE-80, CWE-117).
Claude Opus 5.5 and GPT-6-Sol tied at 73 percent, dethroning GPT-5.5, which led the summer 2026 snapshot at 68 percent. Anthropic and OpenAI each have models in the top seven. In July, six of 11 models sat between 50 and 53 percent; today only two remain in that band, while 10 clear 60 percent. The floor hasn't moved, however, with the bottom still at 44–50 percent, meaning vulnerability volume still depends on model selection.
Data-flow reasoning
Vulnerability type CWE-80, cross-site scripting (XSS), was the main driver of the overall gain. Despite being one of the weaker categories, XSS rose to 18 percent, in contrast to its modest uptick demonstrated in July’s analysis.
This represents the single biggest shift in the dataset. Meanwhile, crypto (CWE-327) held at 87 percent, Structured Query Language (SQL) injection (CWE-89) held at 84 percent, and log injection (CWE-117) remained flat at 13 percent. The reason is structural: XSS is the easiest vulnerability to fix through pattern recognition, which is exactly how LLMs improve. Log injection requires data-flow reasoning across multiple functions, where LLMs continue to stall.
Highest security outcome
The top-performing models carry premium compute costs. Performance also correlates with model size: large models average 55 percent versus 51 percent for small models, which reverses last year's finding that size had "essentially no effect" on security.
“The models that produce the most secure code are the ones that cost the most to run at scale, yet they still fail on more than one in four security tasks,” Wysopal said. “The answer should never be to use the most expensive model and hope for the best. A mid-tier model at a 60 percent pass rate plus Veracode scanning is more accurate, more cost-effective, and more scalable than a frontier model at 73 percent with no verification. Organisations must look for a combination of coding model, security testing, and remediation to get the highest security outcome.”
Coding-specialised models
General models pull further ahead of coding models. General-purpose models average 54 percent on security tasks versus 51 percent for coding-specialised models — a gap that has widened from one to three points since July.
Nearly four in 10 AI code tasks still produce vulnerable code. The rate of vulnerabilities dropped from 44 to 38 percent. This is meaningful but still means roughly four in 10 tasks produced vulnerable code without security-specific guidance.
Java is still stuck. Java remains the dramatic outlier at 31 percent, barely up from 30 percent in July. The gap between Java and the other three languages (Python at 64 percent, C# at 60 percent, and JavaScript at 57 percent) is still 26–33 percentage points.
Long-lived vulnerabilities
The improvement is real but mends a small part of a larger wound. According to Veracode's 2026 State of Software Security report, 82 percent of organisations now carry security debt (vulnerabilities unresolved for more than a year). Moreover, 66 percent of the most dangerous, long-lived vulnerabilities come from third-party code, taking 358 days to fix on average.
“Newer LLMs are producing more code that’s both functional and secure, but that does not mean we can start blindly trusting AI-generated code,” Wysopal closed. “The most effective strategy isn't chasing the most expensive model for marginal gains; it's combining AI code generation with continuous, independent security verification that catches what the models miss. That approach covers the full risk surface, from first-party code to third-party dependencies."
