Analyzing Reddit discussions, the paper finds that while security practitioners see LLMs as useful for boosting productivity, their adoption is constrained by concerns over reliability, verification,…

The paper empirically evaluates domain-adapted and general-purpose LLMs for structured threat modelling (STRIDE on 5G security), finding that domain adaptation and model size do not guarantee reliable…

The paper argues that despite the focus on risk, the cybersecurity profession is structurally trained as a threat-management discipline, leading to poor foundational risk reasoning among professionals…

The paper establishes a standardized security assessment framework and develops a multi-layered defensive system, demonstrating that systematic testing and external defenses are crucial for safe LLM d…

The paper proposes an LLM-enhanced methodology using RAG to automate the creation of security profiles, ensuring compliance with Ukrainian cybersecurity regulations and international best practices.

This paper investigates the practical barriers preventing the trustworthy deployment of AI-driven Cyber Threat Intelligence (CTI) in the highly regulated financial sector, identifying four key socio-t…

The paper introduces a novel six-agent AI architecture for cybersecurity risk assessment, demonstrating high accuracy and speed compared to human experts, though its performance is ultimately limited…

This paper demonstrates that an off-the-shelf Large Language Model (LLM) can function as a high-performing, explainable, human-in-the-loop layer for detecting cyberattacks in Industrial Control System…

The paper introduces ASTRAL, a multimodal LLM-driven framework that reconstructs and analyzes fragmented cyber-physical system architectures to enable comprehensive and quantitative security risk asse…

The paper introduces CyberCertBench, a new benchmark suite for evaluating LLMs against industry cybersecurity certifications, finding that while frontier models perform well on general knowledge, thei…

The paper proposes FinSec, a novel four-tier security detection framework, to robustly identify complex financial risks and suspicious dialogue patterns in LLM-powered financial agents, achieving stat…

The paper proposes Ablating Safety, a controlled protocol for removing safety alignment from language models, demonstrating that targeted de-alignment can significantly boost security performance whil…

The paper demonstrates that adopting LLM-based tools in cybersecurity operations requires a sociotechnical, practitioner-centered co-creation approach, which successfully overcame historical adoption…

The paper conducts a preliminary safety evaluation of the open-weight LLM Kimi K2.5, finding that while it is highly capable, it exhibits concerning dual-use risks, particularly regarding CBRNE misuse…

This paper addresses the critical need for trustworthy LLMs in science by proposing a comprehensive, multi-layered defense framework and methodology to evaluate unique scientific vulnerabilities.

The paper introduces ROK-FORTRESS, a novel bilingual, culturally adversarial benchmark that demonstrates that LLM safety behavior in high-stakes scenarios is significantly shaped by the interaction be…

The paper introduces a validated, consensus-labeled prompt bank that separates requests for executable malicious code (weapons) from requests for general harmful security knowledge, providing a more g…

The paper introduces the CAI Dataset, a massive, multi-terabyte corpus of real-world, hands-on cybersecurity LLM trajectories, designed to address the performance bottleneck caused by expert operator…

The paper introduces PROPARAG, an automated framework that autonomously assesses how well organizational cybersecurity policies comply with standard security controls, achieving high F1 scores on real…

The paper introduces STRIDE-AI, a novel threat modeling framework that adapts classical STRIDE for generative AI, successfully reducing the attack success rate of a tested LLM chatbot from 80% to 15%.