AdapTools: Adaptive Tool-Based Indirect Prompt Injection Attacks on Agentic LLMs

The integration of external data services e.g., Model Context Protocol, MCP has made large language model-based agents increasingly powerful for complex task execution. However, this advancement introduces critical security vulnerabilities, particularly indirect prompt injection IPI attacks...

The Semantic Trap: Do Fine-Tuned LLMs Learn Vulnerability Root Cause or Just Functional Pattern?

LLMs demonstrate promising performance in software vulnerability detection after fine-tuning. However, it remains unclear whether these gains reflect a genuine understanding of vulnerability root causes or merely an exploitation of functional patterns. In this paper, we identify a critical failur...

Learning the Wrong Lessons: Syntactic-Domain Spurious Correlations in Language Models

Whitepaper from researchers at MIT, Northeastern University, and Meta. For an LLM to correctly respond to an instruction it must understand both the semantics and the domain i.e., subject area of a given task-instruction pair. However, syntax can also convey implicit information Recent work shows...

SoK: Security Evaluation of Wi-Fi CSI Biometrics: Attacks, Metrics, and Systemic Weaknesses

Wi-Fi Channel State Information CSI has been repeatedly proposed as a biometric modality, often with reports of high accuracy and operational feasibility. However, the field lacks a consolidated understanding of its security properties, adversarial resilience, and methodological consistency. This...

Architectures, Risks, and Adoption: How to Assess and Choose the Right AI-SOC Platform

Scaling the SOC with AI - Why now? Security Operations Centers SOCs are under unprecedented pressure. According to SACR’s AI-SOC Market Landscape 2025 , the average organization now faces around 960 alerts per day , while large enterprises manage more than 3,000 alerts daily from an average of 28...

Evaluating LLM Generated Detection Rules in Cybersecurity

LLMs are increasingly pervasive in the security environment, with limited measures of their effectiveness, which limits trust and usefulness to security practitioners. Here, we present an open-source evaluation framework and benchmark metrics for evaluating LLM-generated cybersecurity rules. The...

MalEval Android Malware Evaluation Framework

This repository contains the source code of MalEval, an evaluation framework for Android malware behavior auditing, focusing on explaining and substantiating malicious behaviors. The framework provides expert-verified reports, curated metadata, and model outputs to enable reproducible evaluation ...

A Survey and Evaluation Framework for Secure DNS Resolution

Since security was not among the original design goals of the Domain Name System herein called Vanilla DNS, many secure DNS schemes have been proposed to enhance the security and privacy of the DNS resolution process. Some proposed schemes aim to replace the existing DNS infrastructure entirely,...

MAD-Spear: a Conformity-Driven Prompt Injection Attack on Multi-Agent Debate Systems

Multi-agent debate MAD systems leverage collaborative interactions among large language models LLMs agents to improve reasoning capabilities. While recent studies have focused on increasing the accuracy and scalability of MAD systems, their security vulnerabilities have received limited attention...

Vulnerability Mitigation System (VMS): LLM Agent and Evaluation Framework for Autonomous Penetration Testing

As the frequency of cyber threats increases, conventional penetration testing is failing to capture the entirety of todays complex environments. To solve this problem, we propose the Vulnerability Mitigation System VMS, a novel agent based on a Large Language Model LLM capable of performing...

Rethinking and Exploring String-Based Malware Family Classification in the Era of LLMs and RAG

Malware Family Classification MFC aims to identify the fine-grained family e.g., GuLoader or BitRAT to which a potential malware sample belongs, in contrast to malware detection or sample classification that predicts only an Yes/No. Accurate family identification can greatly facilitate automated...

DUMB and DUMBer: Is Adversarial Training Worth It in the Real World?

Adversarial examples are small and often imperceptible perturbations crafted to fool machine learning models. These attacks seriously threaten the reliability of deep neural networks, especially in security-sensitive domains. Evasion attacks, a form of adversarial attack where input is modified a...

From LLMs to MLLMs to Agents: a Survey of Emerging Paradigms in Jailbreak Attacks and Defenses within LLM Ecosystem

Large language models LLMs are rapidly evolving from single-modal systems to multimodal LLMs and intelligent agents, significantly expanding their capabilities while introducing increasingly severe security risks. This paper presents a systematic survey of the growing complexity of jailbreak...

Specification and Evaluation of Multi-Agent LLM Systems -- Prototype and Cybersecurity Applications

Recent advancements in LLMs indicate potential for novel applications, e.g., through reasoning capabilities in the latest OpenAI and DeepSeek models. For applying these models in specific domains beyond text generation, LLM-based multi-agent approaches can be utilized that solve complex tasks by...

Chain-Of-Code Collapse: Reasoning Failures in LLMs Via Adversarial Prompting in Code Generation

Large Language Models LLMs have achieved remarkable success in tasks requiring complex reasoning, such as code generation, mathematical problem solving, and algorithmic synthesis -- especially when aided by reasoning tokens and Chain-of-Thought prompting. Yet, a core question remains: do these...

SoK: Evaluating Jailbreak Guardrails for Large Language Models

Large Language Models LLMs have achieved remarkable progress, but their deployment has exposed critical vulnerabilities, particularly to jailbreak attacks that circumvent safety mechanisms. Guardrails--external defense mechanisms that monitor and control LLM interaction--have emerged as a promisi...

Prompt Attacks Reveal Superficial Knowledge Removal in Unlearning Methods

In this work, we show that some machine unlearning methods may fail when subjected to straightforward prompt attacks. We systematically evaluate eight unlearning techniques across three model families, and employ output-based, logit-based, and probe analysis to determine to what extent supposedly...

Rethinking Machine Unlearning in Image Generation Models

With the surge and widespread application of image generation models, data privacy and content safety have become major concerns and attracted great attention from users, service providers, and policymakers. Machine unlearning MU is recognized as a cost-effective and promising means to address...

Poster: FedBlockParadox -- a Framework for Simulating and Securing Decentralized Federated Learning

A significant body of research in decentralized federated learning focuses on combining the privacy-preserving properties of federated learning with the resilience and transparency offered by blockchain-based systems. While these approaches are promising, they often lack flexible tools to evaluat...

Private Transformer Inference in MLaaS: a Survey

Transformer models have revolutionized AI, powering applications like content generation and sentiment analysis. However, their deployment in Machine Learning as a Service MLaaS raises significant privacy concerns, primarily due to the centralized processing of sensitive user data. Private...