arXiv AI September 25, 2026 By Shenao Yan, Shan Jin, Shimaa Ahmed, Sunpreet Singh Arora, Yiwei Cai, Yizhen Wang, Yuan Hong

Detecting Data Poisoning in Code Generation LLMs via Black-Box, Vulnerability-Oriented Scanning

Read the original on arXiv AI →

The paper introduces CodeScan, a black-box, vulnerability-oriented scanning framework designed to detect data poisoning and backdoor attacks in code generation large language models (LLMs). CodeScan operates by analyzing structural similarities across multiple code generations, normalizing them with abstract syntax tree (AST) techniques, and then applying LLM-based vulnerability analysis to identify recurring insecure patterns. Evaluations on 117 models across three architectures and multiple sizes show over 97% detection accuracy with fewer false positives compared to prior methods.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

llms

arXiv Machine Learning

Aug 17

Detecting Contaminated Code-Generation Prompt Batches via Influence Functions

arXiv:2608. 14303v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used for code generation, yet they remain vulnerable to prompts that elicit insecure implementations.

By Francesco Quinzan, Noor Munir, Yishun Lu, Stephen Roberts

llms benchmarks

arXiv Machine Learning

Jun 30

ANVIL: Anomaly-based Vulnerability Identification without Labelled Training Data

arXiv:2408. 16028v4 Announce Type: replace-cross Abstract: Supervised-learning-based vulnerability detectors often fall short due to limited labelled training data.

By Weizhou Wang, Eric Liu, Xiangyu Guo, Xiao Hu, Ilya Grishchenko, David Lie

llms benchmarks

arXiv Machine Learning

Jun 18

OpenAnt: LLM-Powered Vulnerability Discovery Through Code Decomposition, Adversarial Verification, and Dynamic Testing

arXiv:2606. 19149v1 Announce Type: cross Abstract: Automated vulnerability discovery in large codebases remains challenging: traditional static analysis produces high false-positive rates, while dynamic approaches such as fuzzing require substantial infrastructure and often target narrow classes of bugs.

By Nahum Korda, Gadi Evron

llms safety

arXiv Machine Learning

Jun 16

Data-Centric Benchmarking of Exploit Generation in LLMs: Understanding the Impact of Fine-Tuning

arXiv:2606. 15123v1 Announce Type: cross Abstract: We study the task of CVE-conditioned exploit generation, where a model drafts proof-of-concept (PoC) exploits given software vulnerability context.

By Yiwei Chen, Lichi Li, Kai Cheung, Vinny Parla, Ganesh Sundaram

llms fine-tuning benchmarks data-engineering

arXiv AI

Jul 21

Is "Knowing It's Malicious Enough?" Evaluating LLMs for Fine-Grained Malware Behavior Auditing

arXiv:2509. 14335v2 Announce Type: replace-cross Abstract: Automated malware classifiers achieve strong detection performance, but auditing requires more than flagging a sample: analysts must explain malicious behaviors and justify them with code evidence.

By Xinran Zheng, Xingzhi Qian, Yiling He, Shuo Yang, Lorenzo Cavallaro

llms agents benchmarks

arXiv AI

Jul 10

Beware What You Autocomplete: Forensic Attribution of Backdoored Code Completions

arXiv:2607. 08011v1 Announce Type: cross Abstract: Large language models have enabled powerful code completion systems that assist developers by predicting subsequent lines of code.

By Anjun Gao, Yueyang Quan, Zhuqing Liu, Minghong Fang

llms fine-tuning

One email a morning, machine-written

Related stories

Detecting Contaminated Code-Generation Prompt Batches via Influence Functions

ANVIL: Anomaly-based Vulnerability Identification without Labelled Training Data

OpenAnt: LLM-Powered Vulnerability Discovery Through Code Decomposition, Adversarial Verification, and Dynamic Testing

Data-Centric Benchmarking of Exploit Generation in LLMs: Understanding the Impact of Fine-Tuning

Is "Knowing It's Malicious Enough?" Evaluating LLMs for Fine-Grained Malware Behavior Auditing

Beware What You Autocomplete: Forensic Attribution of Backdoored Code Completions