77 matches found
When Good Sounds Go Adversarial: Jailbreaking Audio-Language Models with Benign Inputs
As large language models become increasingly integrated into daily life, audio has emerged as a key interface for human-AI interaction. However, this convenience also introduces new vulnerabilities, making audio a potential attack surface for adversaries. Our research introduces WhisperInject, a...
LMDG: Advancing Lateral Movement Detection through High-Fidelity Dataset Generation
Lateral Movement LM attacks continue to pose a significant threat to enterprise security, enabling adversaries to stealthily compromise critical assets. However, the development and evaluation of LM detection systems are impeded by the absence of realistic, well-labeled datasets. To address this...
FedBAP: Backdoor Defense Via Benign Adversarial Perturbation in Federated Learning
Federated Learning FL enables collaborative model training while preserving data privacy, but it is highly vulnerable to backdoor attacks. Most existing defense methods in FL have limited effectiveness due to their neglect of the model's over-reliance on backdoor triggers, particularly as the...
Exploring the Secondary Risks of Large Language Models
Ensuring the safety and alignment of Large Language Models is a significant challenge with their growing integration into critical applications and societal functions. While prior research has primarily focused on jailbreak attacks, less attention has been given to non-adversarial failures that...
DAVSP: Safety Alignment for Large Vision-Language Models Via Deep Aligned Visual Safety Prompt
Large Vision-Language Models LVLMs have achieved impressive progress across various applications but remain vulnerable to malicious queries that exploit the visual modality. Existing alignment approaches typically fail to resist malicious queries while preserving utility on benign ones effectivel...
Mind the Gap: a Practical Attack on GGUF Quantization
With the increasing size of frontier LLMs, post-training quantization has become the standard for memory-efficient deployment. Recent work has shown that basic rounding-based quantization schemes pose security risks, as they can be exploited to inject malicious behaviors into quantized models tha...
Keeping an Eye on LLM Unlearning: the Hidden Risk and Remedy
Although Large Language Models LLMs have demonstrated impressive capabilities across a wide range of tasks, growing concerns have emerged over the misuse of sensitive, copyrighted, or harmful data during training. To address these concerns, unlearning techniques have been developed to remove the...
CVE-2023-38831
RARLAB WinRAR before 6.23 allows attackers to execute arbitrary code when a user attempts to view a benign file within a ZIP archive. The issue occurs because a ZIP archive may include a benign file such as an ordinary .JPG file and also a folder that has the same name as the benign file, and the...
CTRAP: Embedding Collapse Trap to Safeguard Large Language Models from Harmful Fine-Tuning
Fine-tuning-as-a-service, while commercially successful for Large Language Model LLM providers, exposes models to harmful fine-tuning attacks. As a widely explored defense paradigm against such attacks, unlearning attempts to remove malicious knowledge from LLMs, thereby essentially preventing th...
Silent Leaks: Implicit Knowledge Extraction Attack on RAG Systems through Benign Queries
Retrieval-Augmented Generation RAG systems enhance large language models LLMs by incorporating external knowledge bases, but they are vulnerable to privacy risks from data extraction attacks. Existing extraction methods typically rely on malicious inputs such as prompt injection or jailbreaking,...
Automated Alert Classification and Triage (AACT): an Intelligent System for the Prioritisation of Cybersecurity Alerts
Enterprise networks are growing ever larger with a rapidly expanding attack surface, increasing the volume of security alerts generated from security controls. Security Operations Centre SOC analysts triage these alerts to identify malicious activity, but they struggle with alert fatigue due to t...
Phishing URL Detection Using Bi-LSTM
Phishing attacks threaten online users, often leading to data breaches, financial losses, and identity theft. Traditional phishing detection systems struggle with high false positive rates and are usually limited by the types of attacks they can identify. This paper proposes a deep learning-based...
Zero Day Malware Detection with Alpha: Fast DBI with Transformer Models for Real World Application
The effectiveness of an AI model in accurately classifying novel malware hinges on the quality of the features it is trained on, which in turn depends on the effectiveness of the analysis tool used. Peekaboo, a Dynamic Binary Instrumentation DBI tool, defeats malware evasion techniques to capture...
MAL-2025-191797 Malicious code in multis (PyPI)
--- -= Per source details. Do not edit below this line.=- Source: kam193 d8dd7fcb7e4ce42262ad3ce89ed580a46e9a2f979c4e2c9db668fb374ae452b8 Infostealer with multiple possibilities, but not auto-activating on installation. There are already multiple attempts to publish it, with different...
kernel: bpf: Address KCSAN report on bpf_lru_list
A data race was found in the BPF LRU list implementation. Concurrent access to node-ref without proper memory barriers triggers KCSAN warnings, though the race is benign as the reference value does not need to be precise...
UNC4990 Leverage Hosting Platforms in USB Infection Chain
Summary: UNC4990, a financially motivated threat actor, has been observed targeting organizations in Italy by utilizing weaponized USB drives as an initial infection vector. Additionally, they are employing trusted websites such as Vimeo, GitHub, and Ars Technica to host encoded payloads disguise...
Exploit for Files or Directories Accessible to External Parties in Apache Struts
CVE-2023-50164 A scanning utility and PoC for CVE-2023-50164...
Exploit for Insufficient Verification of Data Authenticity in Rarlab Winrar
CVE-2023-38831 WinRAR RARLabs WinRAR before 6.23 allows att...
CVE-2023-26443
Full-text autocomplete search allows user-provided SQL syntax to be injected to SQL statements. With existing sanitization in place, this can be abused to trigger benign SQL Exceptions but could potentially be escalated to a malicious SQL injection vulnerability. We now properly encode single...
