Towards an Automated Test of LLM Security Knowledge

Large language models LLMs are increasingly used for a range of software, hardware and human-centered security tasks. Consequently, LLM performance on security tasks is an active area of measurement and research, often with a focus on identifying areas in which LLM security "knowledge'' may be...

CryptanalysisBench: Can LLMs Do Cryptanalysis?

Cryptanalysis - the task of finding attacks against cryptographic schemes - sits at the intersection of mathematical reasoning and cybersecurity, two areas where LLMs have advanced fastest. Cryptanalysis represents both a clean testbed for frontier reasoning as practical attacks can be...

Broken Gates: Re-Evaluating Web Bot Defenses in the Age of LLM Agents

LLM-based browser agents are rapidly changing the threat landscape for web security. Unlike traditional automation frameworks that execute predefined scripts, these agents can autonomously navigate websites, reason about page content, and interact with web interfaces using natural-language...

CVE-2026-64143

In the Linux kernel, the following vulnerability has been resolved: platform/x86: uniwill-laptop: Do not enable the charging limit even when forced It seems that on some older models 2020 the battery charging limit can permanently damage the battery. Prevent users from enabling this feature thru...

CVE-2026-64143

CVE-2026-64143 affects the Linux kernel's platform/x86: uniwill-laptop driver. On certain older Uniwill laptop models (~2020), enabling the battery charging limit feature via the force module parameter can cause permanent battery damage . The root cause is that some devices do not properly implem...

EUVD-2026-45828

In the Linux kernel, the following vulnerability has been resolved: platform/x86: uniwill-laptop: Do not enable the charging limit even when forced It seems that on some older models 2020 the battery charging limit can permanently damage the battery. Prevent users from enabling this feature thru...

Measuring and Evaluating the Performance of Generative AI Models for Scam Detection

Online scams continue to cause substantial financial and personal harm. As a result, detection systems based on Large Language Models LLMs have been integrated into security products ranging from email gateways and browser extensions to fraud-monitoring dashboards. As this adoption accelerates, a...

CVE-2026-16211: Concurrent Execution using Shared Resource with Improper Synchronization

A vulnerability was determined in allegro up to bcf65b994ef29fb3fc2e10b660e6288723d5209e. This impacts the function AssetLastHostname.incrementhostname of the file src/ralph/assets/models/assets.py of the component Hostname Allocation Handler. Executing a manipulation of the argument counter can...

CVE-2026-16212: Concurrent Execution using Shared Resource with Improper Synchronization

A vulnerability was identified in awesto django-shop up to 1.2.4. Affected is an unknown function of the file shop/models/inventory.py of the component Purchase Stock Handler. The manipulation leads to race condition. The attack is possible to be carried out remotely. The attack is considered to...

ShadowPickle: Evading Machine Learning Model Scanners Via Stealthy Pickle Deserialization Attacks

Model hosting hubs e.g., Hugging Face are vulnerable to supply chain attacks that enable remote code execution on trusted user environments. Attackers often distribute malicious Pre-trained ML models PTMs via model hubs. In this paper, we present novel attacks against PTMs and model hubs called...

Buffer Overflow

PocketSphinx is vulnerable to Buffer Overflow. The vulnerability is due to insufficient boundary checks while parsing language model headers and the use of unbounded sscanf string fields when loading acoustic models, which allows an attacker to supply malicious model files that trigger stack or...

EUVD-2026-41914

vLLM: Speech-to-text upload size limit is enforced after full UploadFile read...

Evaluating Open-Weight LLMs for Generating Structured Threat Information for Autonomous Vehicle Vulnerabilities

Connected and Autonomous Vehicles CAVs rely on interconnected software and hardware components, including sensors, Electronic Control Units, in-vehicle infotainment systems, and telematics units, where vulnerabilities can compromise assets, users, and vehicle operations. These vulnerabilities are...

The Language of Security: How Prompt Syntax Shapes Secure Code Generation in Open LLMs

Large Language Models LLMs are increasingly used for source code generation despite their outputs often exhibiting security vulnerabilities. Prior work shows that prompt engineering can mitigate such risks, yet 1 they focused on high-level prompting strategies, neglecting recent evidence that...

CVE-2026-44174

Kirby is an open-source content management system. Prior to 4.9.1 and 5.4.1, Kirby did not validate the model attributes that were used in its collection queries, allowing attackers to include arbitrary model methods in their queries. This includes methods with sensitive data such as password...

CVE-2026-59859

Kiota is an OpenAPI based HTTP Client code generator. Prior to 1.29.1 and 1.32.4, Kiota's PHP generator embedded OpenAPI description, default fields, property names, and other schema-derived strings into PHP double-quoted literals through SanitizeDoubleQuote in Writers/StringExtensions.cs without...

CVE-2026-59861 Kiota: Code Generation Literal Injection in Kiota Ruby Generator

Kiota is an OpenAPI based HTTP Client code generator. Prior to 1.29.1 and 1.32.0, Kiota's Ruby generator embedded OpenAPI default fields, property names, and other schema-derived strings through CodeMethodWriter.cs and SanitizeForQuotedLiteral in Writers/StringExtensions.cs into Ruby double-quote...

[SECURITY] Fedora 44 Update: python-tiktoken-0.13.0-2.fc44

tiktoken is a fast BPE tokeniser for use with OpenAI's models...

PT-2026-60480

Name of the Vulnerable Software and Affected Versions Kiota versions prior to 1.32.0 Description The Ruby generator in Kiota fails to properly escape the character when embedding OpenAPI default fields, property names, and other schema-derived strings into Ruby double-quoted literals. This occurs...

CVE-2026-46336: Improper Limitation of a Pathname to a Restricted Directory

Manyfold is an open source, self-hosted web application for managing a collection of 3d models, particularly focused on 3d printing. From 0.96.0 until 0.140.0, authenticated users can rename uploaded files with path traversal sequences because app/models/modelfile.rb uses the user-controlled...