Research
Academic papers, new techniques, benchmarks, and theoretical findings in AI/LLM security.
Academic papers, new techniques, benchmarks, and theoretical findings in AI/LLM security.
2,003 items
This article from Computers & Security, Volume 172 (January 2027), by Qiang Ruan, Zihang Liu, Chunlai Du, Li Ma and Yanhui Guo, proposes an automated method for analyzing exploitability based on state machines. The source text provided contains only the publication metadata and no details on the method's steps, evaluation or findings.
Researchers Ahmed Abduljawad, Hanane Lamaazi, Muhusina Ismail, Omar EL Harrouss, Saed Alrabaee and Elarbi Badidi published MS-Workset in Computers & Security, Volume 172, in January 2027. The paper introduces a dataset of worker misbehavior for tracking workers' behavior in mobile crowdsensing environments.
Song et al. present IDORacle, a template-guided data-access mediation approach for object-level authorization in database-backed applications, published in Computers & Security, Volume 172, in January 2027. The source text provides only the title, publication date, venue and author list, so the research question, method and findings are not stated.
Hussain M.J. Almohri's paper, published January 2027 in Computers & Security (Volume 172), presents Patronus, a statistical decision-making approach for detecting volumetric application-layer DoS attacks. The source text provided contains only the title, publication details and author, so no method details or findings are available to report.
This paper, published in Computers & Security (Volume 172, January 2027) by Iain Burge, Michel Barbeau and Joaquin Garcia-Alfaro, addresses st-connectivity attacks in quantum networks. The source text provided contains only bibliographic metadata, so the research question, method and findings cannot be stated from it.
Researchers Tianbo Lu, Pengfei Zhao, Yingjie Cai, Subatjan Abdulla and Hanrui Chen published work in Computers & Security, Volume 172, in January 2027 on identifying Tor obfuscated traffic using GAN-based methods on imbalanced datasets. The provided source text contains only the publication metadata and no findings, method details or numbers.
Compliance decision intelligence: An explainable clause-aware framework for automated ETSI consumer IoT security evaluation was published in Computers & Security, Volume 172, in January 2027 by Agung Mulyo Widodo, Chien-Erh Weng, Hsing-Chung Chen and Kevin Christian Atmadja Tioatmodjo. The source text provided contains only publication metadata and no abstract, method or findings.
This systematic literature review examines the economic dimensions of cloud security. It was published in January 2027 in Computers & Security, Volume 172, by Evangelia Filiopoulou, Georgia Dede, Spyridon Evangelatos, Thomas Kamalakis, Christos Michalakelis and Mara Nikolaidou. The source text provided contains only publication metadata and no findings or methods.
The source is a journal citation for a paper titled "Hook & Match - Development and Evaluation of a Pair Matching Phishing Card Game," published in ACM Transactions on Privacy and Security, Volume 29, Issue 4, pages 1-30, November 2026. The source text contains no abstract, methods, or findings, so the question, method and results cannot be summarized from it.
Researchers propose logic-embedded physically unclonable functions (Logic-ePUFs), which replace scan flip-flops and buffers in a semi-custom design flow so the cells act as both standard logic and a PUF for secure key generation. The FF-ePUF and BUF-ePUF were fabricated in 28nm and achieve a bit error rate of 178 ppm with area overhead of 332-447 F2/bit for the FF-ePUF. The BUF-ePUF reduces area overhead by 41% compared to conventional methods.
This paper proposes a detection-aware stochastic optimization framework for industrial control systems facing asynchronous, unreliable feedback and adversarial data poisoning. It uses the SWaT dataset to model poisoning behavior and an adaptive acknowledgment (ACK) bundling mechanism, driven by a multi-objective stochastic gradient descent (SGD) algorithm with a residual-based anomaly indicator, to adjust feedback timing. Numerical experiments on an autonomous surface vehicle (ASV) benchmark show closed-loop recovery under AI-driven data poisoning attacks.
Fix: The source describes an adaptive ACK bundling mechanism that regulates feedback timing and adjusts the ACK bundling window online when anomalous behavior is detected, with Lyapunov–Krasovskii (LK)-based conditions established to ensure closed-loop stability. It does not describe a patch, fixed version or configuration fix for a specific product.
IEEE Xplore (Security & AI Journals)RobustJudge is an automated, modular framework that systematically tests how robust LLM-as-a-Judge systems are across datasets, prompt templates, judge models, attacks and defenses. The study covers 15 attack methods and 8 defense strategies across 13 models, and finds that LLM-based judges stay susceptible under both pointwise and pairwise protocols. Robustness is highly sensitive to prompt-template and judge-model choice, and optimization-based attacks with long suffixes can substantially inflate scores from both PAI-Judge variants.
Researchers propose IFSNet, a weakly supervised, prompt-driven framework that enhances image-level privacy in face images by obscuring soft-biometric attributes such as age, gender, and race while preserving identity utility. It needs no per-image sensitive-attribute annotations during training or inference. The authors report that across five face datasets and three identity-labeled datasets, it reduces sensitive-attribute inference while retaining verification performance, measured by equal error rate and false non-match rate, with a favorable privacy-utility trade-off.