SecCodeBench-V2 Technical Report

2026-02-17

AI Summary

This work addresses the lack of publicly available, industry-oriented evaluation benchmarks for assessing the ability of large language models (LLMs) to generate secure code. To this end, we introduce SecCodeBench-V2, a function-level benchmark for secure code generation and repair that spans five programming languages and 22 Common Weakness Enumeration (CWE) vulnerability categories, comprising 98 real-world vulnerability scenarios along with corresponding proof-of-concept (PoC) test cases. SecCodeBench-V2 is the first benchmark to integrate dynamic execution validation, expert review, and an LLM-as-a-judge mechanism. By modeling tasks at the function level and employing Pass@K scoring aggregation, it enables fine-grained, cross-language, and reproducible evaluation of AI programming assistants in terms of both security and functional correctness.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Large Vision Models

Application Category

• Security and Privacy: Large-scale security measurements
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

We introduce SecCodeBench-V2, a publicly released benchmark for evaluating Large Language Model (LLM) copilots' capabilities of generating secure code. SecCodeBench-V2 comprises 98 generation and fix scenarios derived from Alibaba Group's industrial productions, where the underlying security issues span 22 common CWE (Common Weakness Enumeration) categories across five programming languages: Java, C, Python, Go, and Node.js. SecCodeBench-V2 adopts a function-level task formulation: each scenario provides a complete project scaffold and requires the model to implement or patch a designated target function under fixed interfaces and dependencies. For each scenario, SecCodeBench-V2 provides executable proof-of-concept (PoC) test cases for both functional validation and security verification. All test cases are authored and double-reviewed by security experts, ensuring high fidelity, broad coverage, and reliable ground truth. Beyond the benchmark itself, we build a unified evaluation pipeline that assesses models primarily via dynamic execution. For most scenarios, we compile and run model-generated artifacts in isolated environments and execute PoC test cases to validate both functional correctness and security properties. For scenarios where security issues cannot be adjudicated with deterministic test cases, we additionally employ an LLM-as-a-judge oracle. To summarize performance across heterogeneous scenarios and difficulty levels, we design a Pass@K-based scoring protocol with principled aggregation over scenarios and severity, enabling holistic and comparable evaluation across models. Overall, SecCodeBench-V2 provides a rigorous and reproducible foundation for assessing the security posture of AI coding assistants, with results and artifacts released at https://alibaba.github.io/sec-code-bench. The benchmark is publicly available at https://github.com/alibaba/sec-code-bench.

Problem

• secure code generation
• LLM evaluation
• software security
• CWE
• code benchmark

Innovation

• secure code generation
• LLM evaluation benchmark
• dynamic execution
• LLM-as-a-judge
• Pass@K scoring