PT-2025-32323

Name of the Vulnerable Software and Affected Versions ExecuTorch versions prior to commit cea9b23aa8ff78aff92829a466da97461cc7930c Description A group of related buffer overflow vulnerabilities in the loading of ExecuTorch models can cause the runtime to crash and potentially result in code...

PT-2025-32321

Name of the Vulnerable Software and Affected Versions ExecuTorch versions prior to commit ede82493dae6d2d43f8c424e7be4721abe5242be Description A heap buffer overflow exists in the loading of ExecuTorch models, potentially leading to code execution or other adverse outcomes. Recommendations Update...

adamix-gpt2 (>=0.0.1, <=0.0.2), allennlp (>=1.2.0rc1.dev20200930, <=1.2.0rc1.dev20201021) +6 more potentially affected by CVE-2025-5197 via transformers (3.3.1)

transformers PYPI version 3.3.1 is affected by a known vulnerability. The following packages have a transitive dependency on transformers and may be impacted: - adamix-gpt2 =0.0.1, =1.2.0rc1.dev20200930, =1.2.0rc1.dev20201007, =0.0.1, =0.5.0, =0.1.11, =0.1.12 - mc-transformers 0.1.11 Source cves:...

allennlp (>=0.9.1.dev20200415, <=1.0.0rc4.dev20200511), allennlp-models (>=1.0.0rc2, <=1.0.0rc3) +6 more potentially affected by CVE-2025-5197 via transformers (2.8.0)

transformers PYPI version 2.8.0 is affected by a known vulnerability. The following packages have a transitive dependency on transformers and may be impacted: - allennlp =0.9.1.dev20200415, =1.0.0rc2, =0.1.9, =0.0.2, =0.1.0, =0.3.2 Source cves: CVE-2025-5197 Source advisory: OSV:GHSA-9356-575X-2W...

allennlp (>=1.0.0rc4.dev20200513, <=1.0.0rc6.dev20200601), allennlp-models (>=1.0.0rc4, <=1.0.0rc5) potentially affected by CVE-2025-5197 via transformers (2.9.1)

transformers PYPI version 2.9.1 is affected by a known vulnerability. The following packages have a transitive dependency on transformers and may be impacted: - allennlp =1.0.0rc4.dev20200513, =1.0.0rc4, =1.0.0rc5 Source cves: CVE-2025-5197 Source advisory: OSV:GHSA-9356-575X-2W9M...

adaptnlp (>=0.2.0, <=0.2.2), allennlp (1.2.2) +29 more potentially affected by CVE-2025-5197 via transformers (3.5.1)

transformers PYPI version 3.5.1 is affected by a known vulnerability. The following packages have a transitive dependency on transformers and may be impacted: - adaptnlp =0.2.0, =0.0.7, =0.2.0, =0.0.1, =0.0.1, =0.1.8, =0.1.0, =0.0.8, =0.0.9 and more Source cves: CVE-2025-5197 Source advisory:...

Prompt Injection Vulnerability of Consensus Generating Applications in Digital Democracy

Large Language Models LLMs are gaining traction as a method to generate consensus statements and aggregate preferences in digital democracy experiments. Yet, LLMs may introduce critical vulnerabilities in these systems. Here, we explore the impact of prompt-injection attacks targeting consensus...

From Learning to Unlearning: Biomedical Security Protection in Multimodal Large Language Models

The security of biomedical Multimodal Large Language Models MLLMs has attracted increasing attention. However, training samples easily contain private information and incorrect knowledge that are difficult to detect, potentially leading to privacy leakage or erroneous outputs after deployment. An...

AuthPrint: Fingerprinting Generative Models against Malicious Model Providers

Generative models are increasingly adopted in high-stakes domains, yet current deployments offer no mechanisms to verify the origin of model outputs. We address this gap by extending model fingerprinting techniques beyond the traditional collaborative setting to one where the model provider may a...

AWS VDP: AWS | Self Registration Internal LibreChat : Access to internal/proprietary LLMs

Issue Summary A LibreChat endpoint/UI is found to be accessible to the public Internet, with self registration for any non AWS/Amazon Corporate domains enabled, allowing an attacker to use a ChatGPT like UI to access multiple public models Example : Claude with the API access it has enabled, as...

From Legacy to Standard: LLM-Assisted Transformation of Cybersecurity Playbooks into CACAO Format

Existing cybersecurity playbooks are often written in heterogeneous, non-machine-readable formats, which limits their automation and interoperability across Security Orchestration, Automation, and Response platforms. This paper explores the suitability of Large Language Models, combined with Prom...

BadBlocks: Low-Cost and Stealthy Backdoor Attacks Tailored for Text-To-Image Diffusion Models

In recent years,Diffusion models have achieved remarkable progress in the field of image generation.However,recent studies have shown that diffusion models are susceptible to backdoor attacks,in which attackers can manipulate the output by injecting covert triggers such as specific visual pattern...

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...

SUSE CVE-2025-0928

In Juju versions prior to 3.6.8 and 2.9.52, any authenticated controller user was allowed to upload arbitrary agent binaries to any model or to the controller itself, without verifying model membership or requiring explicit permissions. This enabled the distribution of poisoned binaries to new or...

A Survey on Data Security in Large Language Models

Large Language Models LLMs, now a foundation in advancing natural language processing, power applications such as text generation, machine translation, and conversational systems. Despite their transformative potential, these models inherently rely on massive amounts of training data, often...

Can LLMs Effectively Provide Game-Theoretic-Based Scenarios for Cybersecurity?

Game theory has long served as a foundational tool in cybersecurity to test, predict, and design strategic interactions between attackers and defenders. The recent advent of Large Language Models LLMs offers new tools and challenges for the security of computer systems; In this work, we investiga...

PentestJudge: Judging Agent Behavior against Operational Requirements

We introduce PentestJudge, a system for evaluating the operations of penetration testing agents. PentestJudge is a large language model LLM-as-judge with access to tools that allow it to consume arbitrary trajectories of agent states and tool call history to determine whether a security agent's...

Draytek多款产品 安全漏洞

Draytek AP903 and others are a wireless access point from China Draytek Draytek. A security vulnerability exists in various Draytek products, which stems from an insecure configuration that could lead to unauthorized control of the routing daemon. The following products and versions are affected:...

Large Reasoning Models Are Autonomous Jailbreak Agents

Jailbreaking -- bypassing built-in safety mechanisms in AI models -- has traditionally required complex technical procedures or specialized human expertise. In this study, we show that the persuasive capabilities of large reasoning models LRMs simplify and scale jailbreaking, converting it into a...

Proactive Disentangled Modeling of Trigger-Object Pairings for Backdoor Defense

Deep neural networks DNNs and generative AI GenAI are increasingly vulnerable to backdoor attacks, where adversaries embed triggers into inputs to cause models to misclassify or misinterpret target labels. Beyond traditional single-trigger scenarios, attackers may inject multiple triggers across...