Bias Amplification in RAG: Poisoning Knowledge Retrieval to Steer LLMs

In Large Language Models, Retrieval-Augmented Generation RAG systems can significantly enhance the performance of large language models by integrating external knowledge. However, RAG also introduces new security risks. Existing research focuses mainly on how poisoning attacks in RAG systems affe...

CVE-2025-49653

Exposure of sensitive data in active sessions in Lablup's BackendAI allows attackers to retrieve credentials for users on the management platform...

LLMail-Inject: a Dataset from a Realistic Adaptive Prompt Injection Challenge

Indirect Prompt Injection attacks exploit the inherent limitation of Large Language Models LLMs to distinguish between instructions and data in their inputs. Despite numerous defense proposals, the systematic evaluation against adaptive adversaries remains limited, even when successful attacks ca...

AURA: a Multi-Agent Intelligence Framework for Knowledge-Enhanced Cyber Threat Attribution

Effective attribution of Advanced Persistent Threats APTs increasingly hinges on the ability to correlate behavioral patterns and reason over complex, varied threat intelligence artifacts. We present AURA Attribution Using Retrieval-Augmented Agents, a multi-agent, knowledge-enhanced framework fo...

Safeguarding Multimodal Knowledge Copyright in the RAG-As-A-Service Environment

As Retrieval-Augmented Generation RAG evolves into service-oriented platforms Rag-as-a-Service with shared knowledge bases, protecting the copyright of contributed data becomes essential. Existing watermarking methods in RAG focus solely on textual knowledge, leaving image knowledge unprotected. ...

The vulnerability of the get_file_from_qtn() function in the router_command.sh script of the Quantenna Wi-Fi chip’s software allows a hacker to execute arbitrary commands.

The vulnerability of the getfilefromqtn function in the routercommand.sh script of the Quantenna Wi-Fi chip’s microprogramming system is related to the implementation or modification of arguments. Exploiting this vulnerability could allow a perpetrator to execute arbitrary commands...

Improperly Implemented Security Check for Standard

Overview Affected versions of this package are vulnerable to Improperly Implemented Security Check for Standard due to the exposure of sensitive data in active sessions. An attacker can retrieve credentials for users on the management platform by exploiting this vulnerability. Note: The maintaine...

CVE-2025-49653

CVE-2025-49653 concerns Lablup’s BackendAI. The vulnerability arises from exposure of sensitive data in active sessions, enabling an attacker to retrieve credentials for users on the management platform. The affected software is BackendAI (backend.ai) and specifically involves credentials exposur...

Attacking Attention of Foundation Models Disrupts Downstream Tasks

Foundation models represent the most prominent and recent paradigm shift in artificial intelligence. Foundation models are large models, trained on broad data that deliver high accuracy in many downstream tasks, often without fine-tuning. For this reason, models such as CLIP , DINO or Vision...

CVE-2025-27247

in OpenHarmony v5.0.3 and prior versions allow a local attacker cause information leak through get permission...

SecEmb: Sparsity-Aware Secure Federated Learning of On-Device Recommender System with Large Embedding

Federated recommender system FedRec has emerged as a solution to protect user data through collaborative training techniques. A typical FedRec involves transmitting the full model and entire weight updates between edge devices and the server, causing significant burdens to devices with limited...

CVE-2025-49419

CVE-2025-49419 pertains to Foxit eSign for WordPress. The connected sources describe an Exposure of Sensitive System Information to an Unauthorized Control Sphere vulnerability, enabling retrieval of embedded sensitive data. Affected product: Foxit eSign for WordPress, versions from n/a through 2...

On Automating Security Policies with Contemporary LLMs

The complexity of modern computing environments and the growing sophistication of cyber threats necessitate a more robust, adaptive, and automated approach to security enforcement. In this paper, we present a framework leveraging large language models LLMs for automating attack mitigation policy...

CVE-2025-5598 WF Steuerungstechnik GmbH - airleader MASTER - Path Traversal

Path Traversal vulnerability in WF Steuerungstechnik GmbH airleader MASTER allows Retrieve Embedded Sensitive Data.This issue affects airleader MASTER: 3.0046...

CVE-2025-5598 WF Steuerungstechnik GmbH - airleader MASTER - Path Traversal

Path Traversal vulnerability in WF Steuerungstechnik GmbH airleader MASTER allows Retrieve Embedded Sensitive Data.This issue affects airleader MASTER: 3.0046...

BIT-MOODLE-2024-45689 Moodle: unprotected access to sensitive information via dynamic tables

A flaw was found in Moodle. Dynamic tables did not enforce capability checks, which resulted in users having the ability to retrieve information they did not have permission to access...

Decentralized COVID-19 Health System Leveraging Blockchain

With the development of the Internet, the amount of data generated by the medical industry each year has grown exponentially. The Electronic Health Record EHR manages the electronic data generated during the user's treatment process. Typically, an EHR data manager belongs to a medical institution...

Dot 跨站脚本漏洞

Dot is a text-to-speech, RAG and LLM tool by alexpinel individual developers. A cross-site scripting vulnerability exists in Dot 0.9.3 and earlier versions, which stems from user input and LLM output being appended to the DOM using innerHTML, which could lead to cross-site scripting and command...

CVE-2025-48331

Insertion of Sensitive Information Into Sent Data vulnerability in vanquish WooCommerce Orders & Customers Exporter woocommerce-orders-customers-exporter allows Retrieve Embedded Sensitive Data.This issue affects WooCommerce Orders & Customers Exporter: from n/a through = 5.0...

CVE-2025-48381

Computer Vision Annotation Tool CVAT is an interactive video and image annotation tool for computer vision. In versions starting from 2.4.0 to before 2.38.0, an authenticated CVAT user may be able to retrieve the IDs and names of all tasks, projects, labels, and the IDs of all jobs and quality...