arXiv subjects

Pietro Liguori

Publications and source records attributed to Pietro Liguori.

At least 19 recordsLinked to original sources

arXiv2026

What is the Difference Between Me and You? Benchmarking the Quality Gap Between Human-Written and AI-Generated Code

AI coding assistants are becoming co-authors of production software, yet their evaluation centers on functional correctness, leaving open whether their code differs from human code in the quality dimensions dominating lifecycle cost. We compare human-written and AI-generated code at scale: 787,562 function pairs across Python, Java, and C, each human function mined from open-source repositories paired with implementations generated from its docstring by three AI assistants (OpenAI GPT models, DeepSeek-Coder, Qwen2.5-Coder). We characterize structural complexity and statistical naturalness, and map static-analysis findings onto Orthogonal Defect Classification for defects and the Common Weakness Enumeration for vulnerabilities, making authors and languages directly comparable. AI-generated code is structurally compressed and stylistically templated: roughly half the size and branching of human code, clustering apart at the style level. Defect profiles differ in kind: human code concentrates issues of mature codebases, AI code repetitive boilerplate; security is language-dependent, with LLMs producing more, and more severe, findings in Python and Java but fewer high-severity memory-safety findings than humans in C. Once size is controlled for, complexity metrics carry little signal, while naturalness separates authors. Finally, we release CQBench, a benchmark of 27,346 issue-prone tasks with baselines and an evaluation pipeline for quality assurance and security testing.

cs.SE↗

arXiv2026

Beyond Fixed Fault Models: Comparing LLM-Based and Rule-Based Fault Injection in OpenStack

Software Fault Injection (SFI) supports testing of cloud systems by introducing software defects and observing their manifestation. Rule-based injectors such as ProFIPy provide controlled and reproducible source-level mutations but require fault patterns to be encoded manually. Large Language Models (LLMs) offer a data-driven alternative by generating context-dependent software faults. We compare two code LLMs, Qwen2.5-Coder and DeepSeek-Coder, with ProFIPy in OpenStack's Nova and Cinder services. On shared injection targets, activation and observable-failure rates are comparable, but operational profiles differ: LLM-generated faults produce more Catastrophic outcomes on Nova, whereas ProFIPy produces more Silent and Multi-component effects. The sampled LLM outputs also differ in how they manifest failure, while showing greater agreement in their propagation scope. These findings show that LLM-based fault injection extends the behavioral coverage of fixed fault models without establishing general superiority, and that practical adoption still requires controlled generation, runtime validation, system-level oracles, and reproducible experimental provenance.

cs.SE↗

arXiv2026

AI-Generated PowerShell Malware: An Experimental Framework and Dataset

Generative AI has emerged as a significant cybersecurity threat, with several recent attack campaigns leveraging LLMs to generate code for malicious purposes via scripting languages such as PowerShell. Consequently, for cybersecurity analysts, it is imperative to investigate the offensive capabilities of AI code generators. In this paper, we propose an experimental framework to assess LLM-generated PowerShell malware, which comprises a novel sandbox approach for dynamic analysis of AI-generated malware. Furthermore, we present a novel, manually curated dataset of real-world PowerShell malware, annotated in natural language to assist the training and evaluation of LLMs. Finally, this study evaluates permissive, open-weight LLMs adapted to PowerShell malware generation. Our results reveal a high degree of similarity between real malware and LLM-generated ones in terms of triggered OS malicious events, with a median Jaccard index of 84.5% and 48.4% of instances achieving complete overlap.

cs.CR↗

arXiv2026

Willing but Unable: Separating Refusal from Capability in Code LLMs via Abliteration

Producing a labeled vulnerable code at scale is a recurring obstacle for learning-based vulnerability detection: mined corpora carry substantial label noise, and existing LLM-based augmentation propagates these inaccuracies because it transforms vulnerable seeds rather than synthesising vulnerabilities from a specification. A complementary route is to start from safe code and ask an instruction-tuned LLM to inject a specified CWE (which would shift the labeling burden from open-ended detection to bounded binary confirmation) but safety-aligned code LLMs systematically refuse such prompts. This paper is a preliminary feasibility study of abliteration, a low-rank weight edit that orthogonally projects out the refusal direction in the residual stream, as a tool to remove this barrier. We use Python and CWE-89 (SQL injection) as a case study, evaluating the Qwen2.5-Coder-Instruct family at 3B, 7B, and 14B parameters on safe samples drawn from PromSec and SafeCoder, replicated three times per condition. We find that (i) refusal on injection prompts is strongly size- and prompt-context-dependent: the 14B refuses 100% of prompts, the 7B refuses 73% of PromSec but only 5% of SafeCoder, whereas the 3B is essentially never blocked; (ii) abliteration reduces refusal to zero or near-zero across all sizes while leaving syntactic validity above 93%, supporting the view that, in this setting, refusal can be detached from measured code-generation capability; and (iii) the post-abliteration injection rate remains capacity-bound (88-97% on the 14B, 89-90% on the 7B, and 25-48% on the 3B) separating willingness, which abliteration unlocks, from capability, which scales with parameters. Vulnerability verdicts are produced by a three-tool detector ensemble (CodeQL, Semgrep, Bandit) followed by manual adjudication by two authors on detector-positive outputs.

cs.CR↗

arXiv2026

Will It Break in Production? Metric-Driven Prediction of Residual Defects in Python Systems

Python's dynamic nature complicates testing and increases the possibility that some defects evade detection, so an effective fault prediction becomes essential. We examine whether post-release faults can be predicted using modern ML and DL. Using a balanced dataset of over 4,000 labeled faults with 83 product, process, statistical, and Python-specific metrics plus normalized code representations, we conduct cross-project experiments. LLMs and unsupervised models fail to distinguish residual from non-residual faults, while supervised metric-based models (RandomForest, XGBoost, CatBoost) perform far better, yielding a 0.85-0.9 recall and cutting false negatives by an order of magnitude. Process metrics, especially age, churn, and developer-activity, alongside class and file size, consistently prove most predictive. Notably, the Principal Component Analysis shows that metrics and code embeddings occupy distinct regions of the representation space, suggesting that they capture complementary rather than redundant information.

cs.SE↗

arXiv2025

Human-Written vs. AI-Generated Code: A Large-Scale Study of Defects, Vulnerabilities, and Complexity

As AI code assistants become increasingly integrated into software development workflows, understanding how their code compares to human-written programs is critical for ensuring reliability, maintainability, and security. In this paper, we present a large-scale comparison of code authored by human developers and three state-of-the-art LLMs, i.e., ChatGPT, DeepSeek-Coder, and Qwen-Coder, on multiple dimensions of software quality: code defects, security vulnerabilities, and structural complexity. Our evaluation spans over 500k code samples in two widely used languages, Python and Java, classifying defects via Orthogonal Defect Classification and security vulnerabilities using the Common Weakness Enumeration. We find that AI-generated code is generally simpler and more repetitive, yet more prone to unused constructs and hardcoded debugging, while human-written code exhibits greater structural complexity and a higher concentration of maintainability issues. Notably, AI-generated code also contains more high-risk security vulnerabilities. These findings highlight the distinct defect profiles of AI- and human-authored code and underscore the need for specialized quality assurance practices in AI-assisted programming.

cs.SE↗

arXiv2025

CGP-Tuning: Structure-Aware Soft Prompt Tuning for Code Vulnerability Detection

Large language models (LLMs) have been proposed as powerful tools for detecting software vulnerabilities, where task-specific fine-tuning is typically employed to provide vulnerability-specific knowledge to the LLMs. However, existing fine-tuning techniques often treat source code as plain text, losing the graph-based structural information inherent in code. Graph-enhanced soft prompt tuning addresses this by translating the structural information into contextual cues that the LLM can understand. However, current methods are primarily designed for general graph-related tasks and focus more on adjacency information, they fall short in preserving the rich semantic information (e.g., control/data flow) within code graphs. They also fail to ensure computational efficiency while capturing graph-text interactions in their cross-modal alignment module. This paper presents CGP-Tuning, a new code graph-enhanced, structure-aware soft prompt tuning method for vulnerability detection. CGP-Tuning introduces type-aware embeddings to capture the rich semantic information within code graphs, along with an efficient cross-modal alignment module that achieves linear computational costs while incorporating graph-text interactions. It is evaluated on the latest DiverseVul dataset and three advanced open-source code LLMs, CodeLlama, CodeGemma, and Qwen2.5-Coder. Experimental results show that CGP-Tuning delivers model-agnostic improvements and maintains practical inference speed, surpassing the best graph-enhanced soft prompt tuning baseline by an average of four percentage points and outperforming non-tuned zero-shot prompting by 15 percentage points.

cs.SE↗

arXiv2025

PyResBugs: A Dataset of Residual Python Bugs for Natural Language-Driven Fault Injection

This paper presents PyResBugs, a curated dataset of residual bugs, i.e., defects that persist undetected during traditional testing but later surface in production, collected from major Python frameworks. Each bug in the dataset is paired with its corresponding fault-free (fixed) version and annotated with multi-level natural language (NL) descriptions. These NL descriptions enable natural language-driven fault injection, offering a novel approach to simulating real-world faults in software systems. By bridging the gap between software fault injection techniques and real-world representativeness, PyResBugs provides researchers with a high-quality resource for advancing AI-driven automated testing in Python systems.

cs.SE↗

arXiv2025

Quality In, Quality Out: Investigating Training Data's Role in AI Code Generation

Deep Learning-based code generators have seen significant advancements in recent years. Tools such as GitHub Copilot are used by thousands of developers with the main promise of a boost in productivity. However, researchers have recently questioned their impact on code quality showing, for example, that code generated by DL-based tools may be affected by security vulnerabilities. Since DL models are trained on large code corpora, one may conjecture that low-quality code they output is the result of low-quality code they have seen during training. However, there is very little empirical evidence documenting this phenomenon. Indeed, most of previous work look at the frequency with which commercial code generators recommend low-quality code without the possibility of relating this to their training set. We investigate the extent to which low-quality code instances seen during training affect the quality of the code generated at inference time. We start by fine-tuning a pre-trained DL model on a large-scale dataset being representative of those usually adopted in the training of code generators. We show that 4.98% of functions in this dataset exhibit one or more quality issues related to security, maintainability, best practices, etc. We use the fine-tuned model to generate 551k Python functions, showing that 5.85% of them are affected by at least one quality issue. We then remove from the training set the low-quality functions, and use the cleaned dataset to fine-tune a second model which has been used to generate the same 551k Python functions. We show that the model trained on the cleaned dataset exhibits similar performance in terms of functional correctness as compared to the original model while, however, generating a statistically significant lower number of low-quality functions (2.16%). Our study empirically documents the importance of high-quality training data for code generators.

cs.SE↗

arXiv2024

Enhancing Robustness of AI Offensive Code Generators via Data Augmentation

Since manually writing software exploits for offensive security is time-consuming and requires expert knowledge, AI-base code generators are an attractive solution to enhance security analysts' productivity by automatically crafting exploits for security testing. However, the variability in the natural language and technical skills used to describe offensive code poses unique challenges to their robustness and applicability. In this work, we present a method to add perturbations to the code descriptions to create new inputs in natural language (NL) from well-intentioned developers that diverge from the original ones due to the use of new words or because they miss part of them. The goal is to analyze how and to what extent perturbations affect the performance of AI code generators in the context of offensive code. First, we show that perturbed descriptions preserve the semantics of the original, non-perturbed ones. Then, we use the method to assess the robustness of three state-of-the-art code generators against the newly perturbed inputs, showing that the performance of these AI-based solutions is highly affected by perturbations in the NL descriptions. To enhance their robustness, we use the method to perform data augmentation, i.e., to increase the variability and diversity of the NL descriptions in the training data, proving its effectiveness against both perturbed and non-perturbed code descriptions.

cs.LG↗

arXiv2024

Enhancing AI-based Generation of Software Exploits with Contextual Information

This practical experience report explores Neural Machine Translation (NMT) models' capability to generate offensive security code from natural language (NL) descriptions, highlighting the significance of contextual understanding and its impact on model performance. Our study employs a dataset comprising real shellcodes to evaluate the models across various scenarios, including missing information, necessary context, and unnecessary context. The experiments are designed to assess the models' resilience against incomplete descriptions, their proficiency in leveraging context for enhanced accuracy, and their ability to discern irrelevant information. The findings reveal that the introduction of contextual data significantly improves performance. However, the benefits of additional context diminish beyond a certain point, indicating an optimal level of contextual information for model training. Moreover, the models demonstrate an ability to filter out unnecessary context, maintaining high levels of accuracy in the generation of offensive security code. This study paves the way for future research on optimizing context use in AI-driven code generation, particularly for applications requiring a high degree of technical precision such as the generation of offensive code.

cs.SE↗

arXiv2024

DeVAIC: A Tool for Security Assessment of AI-generated Code

Context: AI code generators are revolutionizing code writing and software development, but their training on large datasets, including potentially untrusted source code, raises security concerns. Furthermore, these generators can produce incomplete code snippets that are challenging to evaluate using current solutions. Objective: This research work introduces DeVAIC (Detection of Vulnerabilities in AI-generated Code), a tool to evaluate the security of AI-generated Python code, which overcomes the challenge of examining incomplete code. Method: We followed a methodological approach that involved gathering vulnerable samples, extracting implementation patterns, and creating regular expressions to develop the proposed tool. The implementation of DeVAIC includes a set of detection rules based on regular expressions that cover 35 Common Weakness Enumerations (CWEs) falling under the OWASP Top 10 vulnerability categories. Results: We utilized four popular AI models to generate Python code, which we then used as a foundation to evaluate the effectiveness of our tool. DeVAIC demonstrated a statistically significant difference in its ability to detect security vulnerabilities compared to the state-of-the-art solutions, showing an F1 Score and Accuracy of 94% while maintaining a low computational cost of 0.14 seconds per code snippet, on average. Conclusions: The proposed tool provides a lightweight and efficient solution for vulnerability detection even on incomplete code.

cs.SE↗

arXiv2024

Automating the Correctness Assessment of AI-generated Code for Security Contexts

Evaluating the correctness of code generated by AI is a challenging open problem. In this paper, we propose a fully automated method, named ACCA, to evaluate the correctness of AI-generated code for security purposes. The method uses symbolic execution to assess whether the AI-generated code behaves as a reference implementation. We use ACCA to assess four state-of-the-art models trained to generate security-oriented assembly code and compare the results of the evaluation with different baseline solutions, including output similarity metrics, widely used in the field, and the well-known ChatGPT, the AI-powered language model developed by OpenAI. Our experiments show that our method outperforms the baseline solutions and assesses the correctness of the AI-generated code similar to the human-based evaluation, which is considered the ground truth for the assessment in the field. Moreover, ACCA has a very strong correlation with the human evaluation (Pearson's correlation coefficient r=0.84 on average). Finally, since it is a fully automated solution that does not require any human intervention, the proposed method performs the assessment of every code snippet in ~0.17s on average, which is definitely lower than the average time required by human analysts to manually inspect the code, based on our experience.

cs.SE↗

arXiv2024

The Power of Words: Generating PowerShell Attacks from Natural Language

As the Windows OS stands out as one of the most targeted systems, the PowerShell language has become a key tool for malicious actors and cybersecurity professionals (e.g., for penetration testing). This work explores an uncharted domain in AI code generation by automatically generating offensive PowerShell code from natural language descriptions using Neural Machine Translation (NMT). For training and evaluation purposes, we propose two novel datasets with PowerShell code samples, one with manually curated descriptions in natural language and another code-only dataset for reinforcing the training. We present an extensive evaluation of state-of-the-art NMT models and analyze the generated code both statically and dynamically. Results indicate that tuning NMT using our dataset is effective at generating offensive PowerShell code. Comparative analysis against the most widely used LLM service ChatGPT reveals the specialized strengths of our fine-tuned models.

cs.CR↗

arXiv2024

Neural Fault Injection: Generating Software Faults from Natural Language

Traditional software fault injection methods, while foundational, face limitations in adequately representing real-world faults, offering customization, and requiring significant manual effort and expertise. This paper introduces a novel methodology that harnesses the capabilities of Large Language Models (LLMs) augmented with Reinforcement Learning from Human Feedback (RLHF) to overcome these challenges. The usage of RLHF emphasizes an iterative refinement process, allowing testers to provide feedback on generated faults, which is then used to enhance the LLM's fault generation capabilities, ensuring the generation of fault scenarios that closely mirror actual operational risks. This innovative methodology aims to significantly reduce the manual effort involved in crafting fault scenarios as it allows testers to focus on higher-level testing strategies, hence paving the way to new possibilities for enhancing the dependability of software systems.

cs.SE↗

arXiv2024

Vulnerabilities in AI Code Generators: Exploring Targeted Data Poisoning Attacks

AI-based code generators have become pivotal in assisting developers in writing software starting from natural language (NL). However, they are trained on large amounts of data, often collected from unsanitized online sources (e.g., GitHub, HuggingFace). As a consequence, AI models become an easy target for data poisoning, i.e., an attack that injects malicious samples into the training data to generate vulnerable code. To address this threat, this work investigates the security of AI code generators by devising a targeted data poisoning strategy. We poison the training data by injecting increasing amounts of code containing security vulnerabilities and assess the attack's success on different state-of-the-art models for code generation. Our study shows that AI code generators are vulnerable to even a small amount of poison. Notably, the attack success strongly depends on the model architecture and poisoning rate, whereas it is not influenced by the type of vulnerabilities. Moreover, since the attack does not impact the correctness of code generated by pre-trained models, it is hard to detect. Lastly, our work offers practical insights into understanding and potentially mitigating this threat.

cs.CR↗

arXiv2024

AI Code Generators for Security: Friend or Foe?

Recent advances of artificial intelligence (AI) code generators are opening new opportunities in software security research, including misuse by malicious actors. We review use cases for AI code generators for security and introduce an evaluation benchmark.

cs.CR↗

arXiv2023

Who Evaluates the Evaluators? On Automatic Metrics for Assessing AI-based Offensive Code Generators

AI-based code generators are an emerging solution for automatically writing programs starting from descriptions in natural language, by using deep neural networks (Neural Machine Translation, NMT). In particular, code generators have been used for ethical hacking and offensive security testing by generating proof-of-concept attacks. Unfortunately, the evaluation of code generators still faces several issues. The current practice uses output similarity metrics, i.e., automatic metrics that compute the textual similarity of generated code with ground-truth references. However, it is not clear what metric to use, and which metric is most suitable for specific contexts. This work analyzes a large set of output similarity metrics on offensive code generators. We apply the metrics on two state-of-the-art NMT models using two datasets containing offensive assembly and Python code with their descriptions in the English language. We compare the estimates from the automatic metrics with human evaluation and provide practical insights into their strengths and limitations.

cs.SE↗