Software Vulnerability Detection

Overview

The field studies how to find vulnerable code and locate the causes of weaknesses across functions, repositories, smart contracts, binaries, and software dependencies. Approaches range from static analysis and fuzzing to transformers, graph neural networks, and LLM agents that combine code structure, security knowledge, and execution-based checks. Benchmarks increasingly test realistic project and temporal splits, localization, exploit validation, and repair, because performance on curated datasets may not transfer to production code. A central challenge is balancing recall with false positives while ensuring that reported findings are grounded in executable evidence and generalize across projects, languages, and vulnerability types.

Latest papers

CardsList

153 papers15 papers in the last four weeks, level with the four weeks before. 0.1% of all new papers.

  1. AVIATOR: Towards AI-Agentic Vulnerability Injection Workflow for High-Fidelity, Large-Scale Code Security Dataset

    Aug 28, 2025Amine Lbath, Massih-Reza Amini, Aurelien Delaitre +1Software SecurityLLMs for Cybersecurity

  2. DMind Benchmark: Toward a Holistic Assessment of LLM Capabilities across the Web3 Domain

    Apr 18, 2025Enhao Huang, Pengyu Sun, Shuxun Wang +13LLM EvaluationSoftware Vulnerability Detection

  3. ANVIL: Anomaly-based Vulnerability Identification without Labelled Training Data

    Aug 28, 2024Weizhou Wang, Eric Liu, Xiangyu Guo +3Software Vulnerability DetectionLLMs for Cybersecurity