Semantic-Preserving Adversarial Attacks on LLMs: an Adaptive Greedy Binary Search Approach

Large Language Models LLMs increasingly rely on automatic prompt engineering in graphical user interfaces GUIs to refine user inputs and enhance response accuracy. However, the diversity of user requirements often leads to unintended misinterpretations, where automated optimizations distort...

CoTGuard: Using Chain-Of-Thought Triggering for Copyright Protection in Multi-Agent LLM Systems

As large language models LLMs evolve into autonomous agents capable of collaborative reasoning and task execution, multi-agent LLM systems have emerged as a powerful paradigm for solving complex problems. However, these systems pose new challenges for copyright protection, particularly when...

Strong Membership Inference Attacks on Massive Datasets and (Moderately) Large Language Models

State-of-the-art membership inference attacks MIAs typically require training many reference models, making it difficult to scale these attacks to large pre-trained language models LLMs. As a result, prior research has either relied on weaker attacks that avoid training reference models e.g.,...

Exemplifying Emerging Phishing: QR-Based Browser-In-The-Browser (BiTB) Attack

Lately, cybercriminals constantly formulate productive approaches to exploit individuals. This article exemplifies an innovative attack, namely QR-based Browser-in-The-Browser BiTB, using proficiencies of Large Language Model LLM i.e. Google Gemini. The presented attack is a fusion of two emergin...

CVE-2024-52581

Litestar is an Asynchronous Server Gateway Interface ASGI framework. Prior to version 2.13.0, the multipart form parser shipped with litestar expects the entire request body as a single byte string and there is no default limit for the total size of the request body. This allows an attacker to...

CVE-2024-31994

Mealie is a self hosted recipe manager and meal planner. Prior to 1.4.0, an attacker can point the image request to an arbitrarily large file. Mealie will attempt to retrieve this file in whole. If it can be retrieved, it may be stored on the file system in whole leading to possible disk...

CVE-2024-23386

memory corruption when WiFi display APIs are invoked with large random inputs...

CVE-2023-28097

OpenSIPS is a Session Initiation Protocol SIP server implementation. Prior to versions 3.1.9 and 3.2.6, a malformed SIP message containing a large Content-Length value and a specially crafted Request-URI causes a segmentation fault in OpenSIPS. This issue occurs when a large amount of shared memo...

CVE-2023-41115

An issue was discovered in EnterpriseDB Postgres Advanced Server EPAS before 11.21.32, 12.x before 12.16.20, 13.x before 13.12.16, 14.x before 14.9.0, and 15.x before 15.4.0. When using UTLENCODE, an authenticated user can read any large object, regardless of that user's permissions...

[SECURITY] Fedora 42 Update: zsync-0.6.2-3.fc42

zsync is a file transfer program. It allows you to download a file from a remote server, where you have a copy of an older version of the file on your computer already. zsync downloads only the new parts of the file. It uses the same algorithm as rsync. However, where rsync is designed for...

CVE-2023-22890

SmartBear Zephyr Enterprise through 7.15.0 allows unauthenticated users to upload large files, which could exhaust the local drive space, causing a denial of service condition...

CVE-2023-1311

A vulnerability, which was classified as critical, was found in SourceCodester Friendly Island Pizza Website and Ordering System 1.0. This affects an unknown part of the file large.php of the component GET Parameter Handler. The manipulation of the argument id leads to sql injection. It is possib...

CVE-2023-21640

Memory corruption in Linux when the file upload API is called with parameters having large buffer...

A Critical Evaluation of Defenses against Prompt Injection Attacks

Large Language Models LLMs are vulnerable to prompt injection attacks, and several defenses have recently been proposed, often claiming to mitigate these attacks successfully. However, we argue that existing studies lack a principled approach to evaluating these defenses. In this paper, we argue...

One Model Transfer to All: on Robust Jailbreak Prompts Generation against LLMs

Safety alignment in large language models LLMs is increasingly compromised by jailbreak attacks, which can manipulate these models to generate harmful or unintended content. Investigating these attacks is crucial for uncovering model vulnerabilities. However, many existing jailbreak strategies fa...

Invisible Tokens, Visible Bills: the Urgent Need to Audit Hidden Operations in Opaque LLM Services

Whitepaper called Invisible Tokens, Visible Bills: The Urgent Need To Audit Hidden Operations In Opaque LLM Services...

ACSE-Eval: Can LLMs Threat Model Real-World Cloud Infrastructure?

While Large Language Models have shown promise in cybersecurity applications, their effectiveness in identifying security threats within cloud deployments remains unexplored. This paper introduces AWS Cloud Security Engineering Eval, a novel dataset for evaluating LLMs cloud security threat...

CVE-2022-2406

The legacy Slack import feature in Mattermost version 6.7.0 and earlier fails to properly limit the sizes of imported files, which allows an authenticated attacker to crash the server by importing large files via the Slack import REST API...

CVE-2022-23580

Tensorflow is an Open Source Machine Learning Framework. During shape inference, TensorFlow can allocate a large vector based on a value from a tensor controlled by the user. The fix will be included in TensorFlow 2.8.0. We will also cherrypick this commit on TensorFlow 2.7.1, TensorFlow 2.6.3, a...

CVE-2022-41907

TensorFlow is an open source platform for machine learning. When tf.rawops.ResizeNearestNeighborGrad is given a large size input, it overflows. We have patched the issue in GitHub commit 00c821af032ba9e5f5fa3fe14690c8d28a657624. The fix will be included in TensorFlow 2.11. We will also cherrypick...