DUP: Detection-Guided Unlearning for Backdoor Purification in Language Models

As backdoor attacks become more stealthy and robust, they reveal critical weaknesses in current defense strategies: detection methods often rely on coarse-grained feature statistics, and purification methods typically require full retraining or additional clean models. To address these challenges...

LLM-Assisted Model-Based Fuzzing of Protocol Implementations

Testing network protocol implementations is critical for ensuring the reliability, security, and interoperability of distributed systems. Faults in protocol behavior can lead to vulnerabilities and system failures, especially in real-time and mission-critical applications. A common approach to...

"Energon": Unveiling Transformers from GPU Power and Thermal Side-Channels

Transformers have become the backbone of many Machine Learning ML applications, including language translation, summarization, and computer vision. As these models are increasingly deployed in shared Graphics Processing Unit GPU environments via Machine Learning as a Service MLaaS, concerns aroun...

Semantic Encryption: Secure and Effective Interaction with Cloud-Based Large Language Models Via Semantic Transformation

The increasing adoption of Cloud-based Large Language Models CLLMs has raised significant concerns regarding data privacy during user interactions. While existing approaches primarily focus on encrypting sensitive information, they often overlook the logical structure of user inputs. This oversig...

CVE-2013-10050 D-Link Devices tools_vct.xgi Authenticated RCE

An OS command injection vulnerability exists in multiple D-Link routers confirmed on DIR-300 rev A v1.05 and DIR-615 rev D v4.13 via the authenticated toolsvct.xgi CGI endpoint. The web interface fails to properly sanitize user-supplied input in the pingIp parameter, allowing attackers with valid...

PT-2025-32521 · Linksys · Linksys Re9000 +5

Name of the Vulnerable Software and Affected Versions: Linksys RE6250, RE6300, RE6350, RE6500, RE7000 and RE9000 versions up to 20250801 Description: A vulnerability exists in Linksys range extenders. The issue affects the setDMZ function within the /goform/setDMZ file. Manipulation of the...

PT-2025-32516 · Linksys · Linksys Re7000 +5

Name of the Vulnerable Software and Affected Versions: Linksys RE6250, RE6300, RE6350, RE6500, RE7000 and RE9000 versions up to 20250801 Description: A vulnerability exists in the ipv6cmd function of the /goform/setIpv6 file. Manipulation of the following arguments leads to OS command injection:...

Using LLMs as a reverse engineering sidekick

This research explores how large language models LLMs can complement, rather than replace, the efforts of malware analysts in the complex field of reverse engineering. LLMs may serve as powerful assistants to streamline workflows, enhance efficiency, and provide actionable insights during malware...

PT-2025-32493 · Pypi · Ms-Swift

This appears to be a security vulnerability report describing a remote code execution RCE exploit in the ms-swift framework through malicious pickle deserialization in adapter model files. The vulnerability allows arbitrary command execution when loading specially crafted adapter models from...

SAEL: Leveraging Large Language Models with Adaptive Mixture-Of-Experts for Smart Contract Vulnerability Detection

With the increasing security issues in blockchain, smart contract vulnerability detection has become a research focus. Existing vulnerability detection methods have their limitations: 1 Static analysis methods struggle with complex scenarios. 2 Methods based on specialized pre-trained models...

How Microsoft defends against indirect prompt injection attacks

Summary The growing adoption of large language models LLMs in enterprise workflows has introduced a new class of adversarial techniques: indirect prompt injection. Indirect prompt injection can be used against systems that leverage large language models LLMs to process untrusted data...

Invisible Injections: Exploiting Vision-Language Models through Steganographic Prompt Embedding

Vision-language models VLMs have revolutionized multimodal AI applications but introduce novel security vulnerabilities that remain largely unexplored. We present the first comprehensive study of steganographic prompt injection attacks against VLMs, where malicious instructions are invisibly...

ZIUM: Zero-Shot Intent-Aware Adversarial Attack on Unlearned Models

Machine unlearning MU removes specific data points or concepts from deep learning models to enhance privacy and prevent sensitive content generation. Adversarial prompts can exploit unlearned models to generate content containing removed concepts, posing a significant security risk. However,...

Secure Coding for Web Applications: Frameworks, Challenges, and the Role of LLMs

Secure coding is a critical yet often overlooked practice in software development. Despite extensive awareness efforts, real-world adoption remains inconsistent due to organizational, educational, and technical barriers. This paper provides a comprehensive review of secure coding practices across...

Can We End the Cat-And-Mouse Game? Simulating Self-Evolving Phishing Attacks with LLMs and Genetic Algorithms

Anticipating emerging attack methodologies is crucial for proactive cybersecurity. Recent advances in Large Language Models LLMs have enabled the automated generation of phishing messages and accelerated research into potential attack techniques. However, predicting future threats remains...

Securing Cloud AI and LLMs with TotalAI for Visibility, Risk Context and Control

As enterprises accelerate AI adoption, large language models LLMs hosted on public cloud platforms are quickly becoming the norm due to their simplified access and pricing model. Cloud-native services like AWS Bedrock, Azure AI Foundry, and Google Vertex AI offer powerful, pay-as-you-go access to...

CVE-2025-54413

A flaw was found in skops. An inconsistency in MethodNode allows access to unexpected object fields through dot notation when a specially crafted model file is loaded. This issue allows arbitrary code execution at load time...

Cascading and Proxy Membership Inference Attacks

A Membership Inference Attack MIA assesses how much a trained machine learning model reveals about its training data by determining whether specific query instances were included in the dataset. We classify existing MIAs into adaptive or non-adaptive, depending on whether the adversary is allowed...

LLMs Framework Recursive Reference DoS

The vulnerability of the LlamaIndex framework for working with large language models is related to an improper limitation on recursive references to entities in the DTD. Exploiting this vulnerability could allow a malicious actor to cause service failures...

Ollama Code Execution

The vulnerability of Ollama’s system for running and managing large language models is related to insufficient validation of input data. Exploiting this vulnerability could allow a remote attacker to execute arbitrary code...