Although Deep Learning (DL) methods becoming increasingly popular in vulnerability detection, their performance is seriously limited by insufficient training data. This is mainly because few existing software organizations can maintain a complete set of high-quality samples for DL-based vulnerability detection. Due to the concerns about privacy leakage, most of them are reluctant to share data, resulting in the data silo problem. Since enables collaboratively model training without data sharing, Federated Learning (FL) has been investigated as a promising means of addressing the data silo problem in DL-based vulnerability detection. However, since existing FL-based vulnerability detection methods focus on specific applications, it is still far unclear i) how well FL adapts to common vulnerability detection tasks and ii) how to design a high-performance FL solution for a specific vulnerability detection task. To answer these two questions, this paper first proposes VulFL, an effective evaluation framework for FL-based vulnerability detection. Then, based on VulFL, this paper conducts a comprehensive study to reveal the underlying capabilities of FL in dealing with different types of CWEs, especially when facing various data heterogeneity scenarios. Our experimental results show that, compared to independent training, FL can significantly improve the detection performance of common AI models on all investigated CWEs, though the performance of FL-based vulnerability detection is limited by heterogeneous data. To highlight the performance differences between different FL solutions for vulnerability detection, we extensively investigate the impacts of different configuration strategies for each framework component of VulFL. Our study sheds light on the potential of FL in vulnerability detection, which can be used to guide the design of FL-based solutions for vulnerability detection.
Cryptographic protocols play a fundamental role in securing modern digital infrastructure, but they are often deployed without prior formal verification. This could lead to the adoption of distributed systems vulnerable to attack vectors. Formal verification methods, on the other hand, require complex and time-consuming techniques that lack automatization. In this paper, we introduce a benchmark to assess the ability of Large Language Models (LLMs) to autonomously identify vulnerabilities in new cryptographic protocols through interaction with Tamarin: a theorem prover for protocol verification. We created a manually validated dataset of novel, flawed, communication protocols and designed a method to automatically verify the vulnerabilities found by the AI agents. Our results about the performances of the current frontier models on the benchmark provides insights about the possibility of cybersecurity applications by integrating LLMs with symbolic reasoning systems.
With the rapid development of blockchain technology, smart contract security has become a critical challenge. Existing smart contract vulnerability detection methods face three main issues: (1) Insufficient quality of datasets, lacking detailed explanations and precise vulnerability locations. (2) Limited adaptability of large language models (LLMs) to the smart contract domain, as most LLMs are pre-trained on general text data but minimal smart contract-specific data. (3) Lack of high-quality explanations for detected vulnerabilities, as existing methods focus solely on detection without clear explanations. These limitations hinder detection performance and make it harder for developers to understand and fix vulnerabilities quickly, potentially leading to severe financial losses. To address these problems, we propose Smart-LLaMA, an advanced detection method based on the LLaMA language model. First, we construct a comprehensive dataset covering four vulnerability types with labels, detailed explanations, and precise vulnerability locations. Second, we introduce Smart Contract-Specific Continual Pre-Training, using raw smart contract data to enable the LLM to learn smart contract syntax and semantics, enhancing their domain adaptability. Furthermore, we propose Explanation-Guided Fine-Tuning, which fine-tunes the LLM using paired vulnerable code and explanations, enabling both vulnerability detection and reasoned explanations. We evaluate explanation quality through LLM and human evaluation, focusing on Correctness, Completeness, and Conciseness. Experimental results show that Smart-LLaMA outperforms state-of-the-art baselines, with average improvements of 6.49% in F1 score and 3.78% in accuracy, while providing reliable explanations.
This paper presents LProtector, an automated vulnerability detection system for C/C++ codebases driven by the large language model (LLM) GPT-4o and Retrieval-Augmented Generation (RAG). As software complexity grows, traditional methods face challenges in detecting vulnerabilities effectively. LProtector leverages GPT-4o's powerful code comprehension and generation capabilities to perform binary classification and identify vulnerabilities within target codebases. We conducted experiments on the Big-Vul dataset, showing that LProtector outperforms two state-of-the-art baselines in terms of F1 score, demonstrating the potential of integrating LLMs with vulnerability detection.
Binary analysis is a core component of many critical security tasks, including reverse engineering, malware analysis, and vulnerability detection. Manual analysis is often time-consuming, but identifying commonly-used or previously-seen functions can reduce the time it takes to understand a new file. However, given the complexity of assembly, and the NP-hard nature of determining function equivalence, this task is extremely difficult. Common approaches often use sophisticated disassembly and decompilation tools, graph analysis, and other expensive pre-processing steps to perform function similarity searches over some corpus. In this work, we identify a number of discrepancies between the current research environment and the underlying application need. To remedy this, we build a new benchmark, REFuSE-Bench, for binary function similarity detection consisting of high-quality datasets and tests that better reflect real-world use cases. In doing so, we address issues like data duplication and accurate labeling, experiment with real malware, and perform the first serious evaluation of ML binary function similarity models on Windows data. Our benchmark reveals that a new, simple basline, one which looks at only the raw bytes of a function, and requires no disassembly or other pre-processing, is able to achieve state-of-the-art performance in multiple settings. Our findings challenge conventional assumptions that complex models with highly-engineered features are being used to their full potential, and demonstrate that simpler approaches can provide significant value.
Our work explores the utilization of deep learning, specifically leveraging the CodeBERT model, to enhance code security testing for Python applications by detecting SQL injection vulnerabilities. Unlike traditional security testing methods that may be slow and error-prone, our approach transforms source code into vector representations and trains a Long Short-Term Memory (LSTM) model to identify vulnerable patterns. When compared with existing static application security testing (SAST) tools, our model displays superior performance, achieving higher precision, recall, and F1-score. The study demonstrates that deep learning techniques, particularly with CodeBERT's advanced contextual understanding, can significantly improve vulnerability detection, presenting a scalable methodology applicable to various programming languages and vulnerability types.
We introduce a new benchmark for assessing AI models' capabilities and risks in automated software exploitation, focusing on their ability to detect and exploit vulnerabilities in real-world software systems. Using DARPA's AI Cyber Challenge (AIxCC) framework and the Nginx challenge project, a deliberately modified version of the widely used Nginx web server, we evaluate several leading language models, including OpenAI's o1-preview and o1-mini, Anthropic's Claude-3.5-sonnet-20241022 and Claude-3.5-sonnet-20240620, Google DeepMind's Gemini-1.5-pro, and OpenAI's earlier GPT-4o model. Our findings reveal that these models vary significantly in their success rates and efficiency, with o1-preview achieving the highest success rate of 64.71 percent and o1-mini and Claude-3.5-sonnet-20241022 providing cost-effective but less successful alternatives. This benchmark establishes a foundation for systematically evaluating the AI cyber risk posed by automated exploitation tools.
The latest advancements in large language models (LLMs) have sparked interest in their potential for software vulnerability detection. However, there is currently a lack of research specifically focused on vulnerabilities in the PHP language, and challenges in extracting samples and processing persist, hindering the model's ability to effectively capture the characteristics of specific vulnerabilities. In this paper, we present RealVul, the first LLM-based framework designed for PHP vulnerability detection, addressing these issues. By vulnerability candidate detection methods and employing techniques such as normalization, we can isolate potential vulnerability triggers while streamlining the code and eliminating unnecessary semantic information, enabling the model to better understand and learn from the generated vulnerability samples. We also address the issue of insufficient PHP vulnerability samples by improving data synthesis methods. To evaluate RealVul's performance, we conduct an extensive analysis using five distinct code LLMs on vulnerability data from 180 PHP projects. The results demonstrate a significant improvement in both effectiveness and generalization compared to existing methods, effectively boosting the vulnerability detection capabilities of these models.
Vulnerability detection is a crucial yet challenging task to identify potential weaknesses in software for cyber security. Recently, deep learning (DL) has made great progress in automating the detection process. Due to the complex multi-layer structure and a large number of parameters, a DL model requires massive labeled (vulnerable or secure) source code to gain knowledge to effectively distinguish between vulnerable and secure code. In the literature, many datasets have been created to train DL models for this purpose. However, these datasets suffer from several issues that will lead to low detection accuracy of DL models. In this paper, we define three critical issues (i.e., data imbalance, low vulnerability coverage, biased vulnerability distribution) that can significantly affect the model performance and three secondary issues (i.e., errors in source code, mislabeling, noisy historical data) that also affect the performance but can be addressed through a dedicated pre-processing procedure. In addition, we conduct a study of 14 papers along with 54 datasets for vulnerability detection to confirm these defined issues. Furthermore, we discuss good practices to use existing datasets and to create new ones.
Recently, deep learning (DL) approaches to vulnerability detection have gained significant traction. These methods demonstrate promising results, often surpassing traditional static code analysis tools in effectiveness. In this study, we explore a novel approach to vulnerability detection utilizing the tools from topological data analysis (TDA) on the attention matrices of the BERT model. Our findings reveal that traditional machine learning (ML) techniques, when trained on the topological features extracted from these attention matrices, can perform competitively with pre-trained language models (LLMs) such as CodeBERTa. This suggests that TDA tools, including persistent homology, are capable of effectively capturing semantic information critical for identifying vulnerabilities.
Large Language Models (LLMs) have shown significant challenges in detecting and repairing vulnerable code, particularly when dealing with vulnerabilities involving multiple aspects, such as variables, code flows, and code structures. In this study, we utilize GitHub Copilot as the LLM and focus on buffer overflow vulnerabilities. Our experiments reveal a notable gap in Copilot's abilities when dealing with buffer overflow vulnerabilities, with a 76% vulnerability detection rate but only a 15% vulnerability repair rate. To address this issue, we propose context-aware prompt tuning techniques designed to enhance LLM performance in repairing buffer overflow. By injecting a sequence of domain knowledge about the vulnerability, including various security and code contexts, we demonstrate that Copilot's successful repair rate increases to 63%, representing more than four times the improvement compared to repairs without domain knowledge.
Large Language Models (LLMs) have shown promise in tasks like code translation, prompting interest in their potential for automating software vulnerability detection (SVD) and patching (SVP). To further research in this area, establishing a benchmark is essential for evaluating the strengths and limitations of LLMs in these tasks. Despite their capabilities, questions remain regarding whether LLMs can accurately analyze complex vulnerabilities and generate appropriate patches. This paper introduces VulnLLMEval, a framework designed to assess the performance of LLMs in identifying and patching vulnerabilities in C code. Our study includes 307 real-world vulnerabilities extracted from the Linux kernel, creating a well-curated dataset that includes both vulnerable and patched code. This dataset, based on real-world code, provides a diverse and representative testbed for evaluating LLM performance in SVD and SVP tasks, offering a robust foundation for rigorous assessment. Our results reveal that LLMs often struggle with distinguishing between vulnerable and patched code. Furthermore, in SVP tasks, these models tend to oversimplify the code, producing solutions that may not be directly usable without further refinement.
The large-scale deployment of Solidity smart contracts on the Ethereum mainnet has increasingly attracted financially-motivated attackers in recent years. A few now-infamous attacks in Ethereum's history includes DAO attack in 2016 (50 million dollars lost), Parity Wallet hack in 2017 (146 million dollars locked), Beautychain's token BEC in 2018 (900 million dollars market value fell to 0), and NFT gaming blockchain breach in 2022 ($600 million in Ether stolen). This paper presents a comprehensive investigation of the use of large language models (LLMs) and their capabilities in detecting OWASP Top Ten vulnerabilities in Solidity. We introduce a novel, class-balanced, structured, and labeled dataset named VulSmart, which we use to benchmark and compare the performance of open-source LLMs such as CodeLlama, Llama2, CodeT5 and Falcon, alongside closed-source models like GPT-3.5 Turbo and GPT-4o Mini. Our proposed SmartVD framework is rigorously tested against these models through extensive automated and manual evaluations, utilizing BLEU and ROUGE metrics to assess the effectiveness of vulnerability detection in smart contracts. We also explore three distinct prompting strategies-zero-shot, few-shot, and chain-of-thought-to evaluate the multi-class classification and generative capabilities of the SmartVD framework. Our findings reveal that SmartVD outperforms its open-source counterparts and even exceeds the performance of closed-source base models like GPT-3.5 and GPT-4 Mini. After fine-tuning, the closed-source models, GPT-3.5 Turbo and GPT-4o Mini, achieved remarkable performance with 99% accuracy in detecting vulnerabilities, 94% in identifying their types, and 98% in determining severity. Notably, SmartVD performs best with the `chain-of-thought' prompting technique, whereas the fine-tuned closed-source models excel with the `zero-shot' prompting approach.
Supervised-learning-based vulnerability detectors often fall short due to limited labelled training data. In contrast, Large Language Models (LLMs) like GPT-4 are trained on vast unlabelled code corpora, yet perform only marginally better than coin flips when directly prompted to detect vulnerabilities. In this paper, we reframe vulnerability detection as anomaly detection, based on the premise that vulnerable code is rare and thus anomalous relative to patterns learned by LLMs. We introduce ANVIL, which performs a masked code reconstruction task: the LLM reconstructs a masked line of code, and deviations from the original are scored as anomalies. We propose a hybrid anomaly score that combines exact match, cross-entropy loss, prediction confidence, and structural complexity. We evaluate our approach across multiple LLM families, scoring methods, and context sizes, and against vulnerabilities after the LLM's training cut-off. On the PrimeVul dataset, ANVIL outperforms state-of-the-art supervised detectors-LineVul, LineVD, and LLMAO-achieving up to 2x higher Top-3 accuracy, 75% better Normalized MFR, and a significant improvement on ROC-AUC. Finally, by integrating ANVIL with fuzzers, we uncover two previously unknown vulnerabilities, demonstrating the practical utility of anomaly-guided detection.
According to our survey of machine learning for vulnerability detection (ML4VD), 9 in every 10 papers published in the past five years define ML4VD as a function-level binary classification problem: Given a function, does it contain a security flaw? From our experience as security researchers, faced with deciding whether a given function makes the program vulnerable to attacks, we would often first want to understand the context in which this function is called. In this paper, we study how often this decision can really be made without further context and study both vulnerable and non-vulnerable functions in the most popular ML4VD datasets. We call a function "vulnerable" if it was involved in a patch of an actual security flaw and confirmed to cause the program's vulnerability. It is "non-vulnerable" otherwise. We find that in almost all cases this decision cannot be made without further context. Vulnerable functions are often vulnerable only because a corresponding vulnerability-inducing calling context exists while non-vulnerable functions would often be vulnerable if a corresponding context existed. But why do ML4VD techniques achieve high scores even though there is demonstrably not enough information in these samples? Spurious correlations: We find that high scores can be achieved even when only word counts are available. This shows that these datasets can be exploited to achieve high scores without actually detecting any security vulnerabilities. We conclude that the prevailing problem statement of ML4VD is ill-defined and call into question the internal validity of this growing body of work. Constructively, we call for more effective benchmarking methodologies to evaluate the true capabilities of ML4VD, propose alternative problem statements, and examine broader implications for the evaluation of machine learning and programming analysis research.
Though many deep learning-based models have made great progress in vulnerability detection, we have no good understanding of these models, which limits the further advancement of model capability, understanding of the mechanism of model detection, and efficiency and safety of practical application of models. In this paper, we extensively and comprehensively investigate two types of state-of-the-art learning-based approaches (sequence-based and graph-based) by conducting experiments on a recently built large-scale dataset. We investigate seven research questions from five dimensions, namely model capabilities, model interpretation, model stability, ease of use of model, and model economy. We experimentally demonstrate the priority of sequence-based models and the limited abilities of both LLM (ChatGPT) and graph-based models. We explore the types of vulnerability that learning-based models skilled in and reveal the instability of the models though the input is subtlely semantical-equivalently changed. We empirically explain what the models have learned. We summarize the pre-processing as well as requirements for easily using the models. Finally, we initially induce the vital information for economically and safely practical usage of these models.
Binary program vulnerability detection is critical for software security, yet existing deep learning approaches often rely on source code analysis, limiting their ability to detect unknown vulnerabilities. To address this, we propose VulCatch, a binary-level vulnerability detection framework. VulCatch introduces a Synergy Decompilation Module (SDM) and Kolmogorov-Arnold Networks (KAN) to transform raw binary code into pseudocode using CodeT5, preserving high-level semantics for deep analysis with tools like Ghidra and IDA. KAN further enhances feature transformation, enabling the detection of complex vulnerabilities. VulCatch employs word2vec, Inception Blocks, BiLSTM Attention, and Residual connections to achieve high detection accuracy (98.88%) and precision (97.92%), while minimizing false positives (1.56%) and false negatives (2.71%) across seven CVE datasets.
Detecting vulnerabilities is vital for software security, yet deep learning-based vulnerability detectors (DLVD) face a data shortage, which limits their effectiveness. Data augmentation can potentially alleviate the data shortage, but augmenting vulnerable code is challenging and requires a generative solution that maintains vulnerability. Previous works have only focused on generating samples that contain single statements or specific types of vulnerabilities. Recently, large language models (LLMs) have been used to solve various code generation and comprehension tasks with inspiring results, especially when fused with retrieval augmented generation (RAG). Therefore, we propose VulScribeR, a novel LLM-based solution that leverages carefully curated prompt templates to augment vulnerable datasets. More specifically, we explore three strategies to augment both single and multi-statement vulnerabilities, with LLMs, namely Mutation, Injection, and Extension. Our extensive evaluation across four vulnerability datasets and DLVD models, using three LLMs, show that our approach beats two SOTA methods Vulgen and VGX, and Random Oversampling (ROS) by 27.48%, 27.93%, and 15.41% in f1-score with 5K generated vulnerable samples on average, and 53.84%, 54.10%, 69.90%, and 40.93% with 15K generated vulnerable samples. Our approach demonstrates its feasibility for large-scale data augmentation by generating 1K samples at as cheap as US$ 1.88.
Software vulnerabilities, caused by unintentional flaws in source code, are a primary root cause of cyberattacks. Static analysis of source code has been widely used to detect these unintentional defects introduced by software developers. Large Language Models (LLMs) have demonstrated human-like conversational abilities due to their capacity to capture complex patterns in sequential data, such as natural languages. In this paper, we harness LLMs' capabilities to analyze source code and detect known vulnerabilities. To ensure the proposed vulnerability detection method is universal across multiple programming languages, we convert source code to LLVM IR and train LLMs on these intermediate representations. We conduct extensive experiments on various LLM architectures and compare their accuracy. Our comprehensive experiments on real-world and synthetic codes from NVD and SARD demonstrate high accuracy in identifying source code vulnerabilities.
With the rise of large language models (LLMs), researchers are increasingly exploring their applications in var ious vertical domains, such as software engineering. LLMs have achieved remarkable success in areas including code generation and vulnerability detection. However, they also exhibit numerous limitations and shortcomings. LLM-based agents, a novel tech nology with the potential for Artificial General Intelligence (AGI), combine LLMs as the core for decision-making and action-taking, addressing some of the inherent limitations of LLMs such as lack of autonomy and self-improvement. Despite numerous studies and surveys exploring the possibility of using LLMs in software engineering, it lacks a clear distinction between LLMs and LLM based agents. It is still in its early stage for a unified standard and benchmarking to qualify an LLM solution as an LLM-based agent in its domain. In this survey, we broadly investigate the current practice and solutions for LLMs and LLM-based agents for software engineering. In particular we summarise six key topics: requirement engineering, code generation, autonomous decision-making, software design, test generation, and software maintenance. We review and differentiate the work of LLMs and LLM-based agents from these six topics, examining their differences and similarities in tasks, benchmarks, and evaluation metrics. Finally, we discuss the models and benchmarks used, providing a comprehensive analysis of their applications and effectiveness in software engineering. We anticipate this work will shed some lights on pushing the boundaries of LLM-based agents in software engineering for future research.