CVE-2026-31239

The mamba language model framework thru 2.2.6 is vulnerable to insecure deserialization CWE-502 when loading pre-trained models from HuggingFace Hub. The MambaLMHeadModel.frompretrained method uses torch.load to load the pytorchmodel.bin weight file without enabling the security-restrictive...

PT-2026-40126

The mamba language model framework thru 2.2.6 is vulnerable to insecure deserialization CWE-502 when loading pre-trained models from HuggingFace Hub. The MambaLMHeadModel.from pretrained method uses torch.load to load the pytorch model.bin weight file without enabling the security-restrictive...

When LLMs Team Up: A Coordinated Attack Framework for Automated Cyber Intrusions

Automated intrusion-style workflows require LLM agents to reason over partial observations, tool outputs, and executable artifacts under bounded budgets. A single LLM instance often compresses evidence extraction, planning, execution, and validation into one context, which increases the risk of...

CVE-2026-31228

The Adversarial Robustness Toolbox ART thru 1.20.1 contains a remote code execution vulnerability in its Kubeflow component. The robustness evaluation function for PyTorch models uses the unsafe eval function to dynamically evaluate user-supplied strings for the LossFn and Optimizer parameters...

Still Camouflage, Moving Illusion: View-Induced Trajectory Manipulation in Autonomous Driving

Existing physical adversarial attacks on vision-based autonomous driving induce time-evolving perception errors, including biased object tracking or trajectory prediction, through i sophisticated physical patch inducing detection box drift when entering the view distance, or ii dynamically changi...

CVE-2026-31239

The CVE-2026-31239 entry concerns the Mamba language model framework up to version 2.2.6. The issue is insecure deserialization (CWE-502) when loading pre-trained models from HuggingFace Hub. The MambaLMHeadModel.from_pretrained() method uses torch.load() to load the pytorch_model.bin weight file...

EUVD-2026-29151

OpenClaw before 2026.4.23 contains an improper access control vulnerability in the gateway tool's config.apply and config.patch operations that allows compromised models to write unsafe configuration changes by bypassing an incomplete denylist protection. Attackers can persist malicious config...

CVE-2026-45006

OpenClaw before 2026.4.23 contains an improper access control vulnerability in the gateway tool's config.apply and config.patch operations that allows compromised models to write unsafe configuration changes by bypassing an incomplete denylist protection. Attackers can persist malicious config...

CVE-2026-45006

CVE-2026-45006 affects OpenClaw prior to 2026.4.23, due to improper access control in the gateway tool’s config.apply and config.patch. The vulnerability bypasses an incomplete denylist, allowing compromised models to persist unsafe configuration changes that can alter command execution, network ...

CVE-2026-7813 pgAdmin 4: Cross-user data access and shared-server privilege escalation in server mode

Authorization vulnerability in pgAdmin 4 server mode affecting Server Groups, Servers, Shared Servers, Background Processes, and Debugger modules. Multiple endpoints fetched user-owned objects without filtering by the requesting user's identity. An authenticated user could access another user's...

LLMs and Text-in-Text Steganography

Turns out that LLMs are really good at hiding text messages in other text messages...

Guaranteed Jailbreaking Defense Via Disrupt-And-Rectify Smoothing

This paper proposes a guaranteed defense method for large language models LLMs to safeguard against jailbreaking attacks. Drawing inspiration from the denoised-smoothing approach in the adversarial defense domain, we propose a novel smoothing-based defense method, termed Disrupt-and-Rectify...

CosyVoice 安全漏洞

CosyVoice is an open-source voice generation and AI voice cloning platform developed by FunAudioLLM. CosyVoice has a security vulnerability, which stems from the gRPC server component using torch.load to load the voice synthesis model without enabling the weights-only=True security parameter. Thi...

LLMs for Secure Hardware Design and Related Problems: Opportunities and Challenges

The integration of Large Language Models LLMs into Electronic Design Automation EDA and hardware security is rapidly reshaping the semiconductor industry. While LLMs offer unprecedented capabilities in generating Register Transfer Level RTL code, automating testbenches, and bridging the semantic...

Adversarial SQL Injection Generation with LLM-Based Architectures

SQL injection SQLi attacks are still one of the serious attacks ranked in the Open Worldwide Application Security Project OWASP Top 10 threats. Today, with advances in Artificial Intelligence AI, especially in Large Language Models LLMs, an opportunity has been created for automating adversarial...

Position: AI Security Policy Should Target Systems, Not Models

We present swarm-attack, an open-source adversarial testing framework in which multiple lightweight LLM agents coordinate through shared memory, parallel exploration, and evolutionary optimization. Together, our results demonstrate that both safety bypass of frontier models and software...

Missing Authorization

Overview open-webui is an Open WebUI Affected versions of this package are vulnerable to Missing Authorization in the generatecompletion, embed, embeddings, and showmodelinfo functions. An attacker can access restricted model information and consume compute resources by sending crafted API reques...

Missing Authorization

Overview open-webui is an Open WebUI Affected versions of this package are vulnerable to Missing Authorization through the importmodels process. An attacker can overwrite existing models owned by other users, modify their configuration, and escalate access by submitting crafted payloads to the...

Open WebUI's Model Import Overwrites Any Model Without Ownership Check

Model Import Overwrites Any Model Without Ownership Check Affected Component Model import endpoint: - backend/openwebui/routers/models.py lines 254-308, importmodels Affected Versions Current main branch commit 6fdd19bf1 and likely all versions with model import functionality. Description The POS...

GHSA-HP5M-24VP-VQ2Q Open WebUI's responses passthrough endpoint lacks access control authorization

Summary The /responses endpoint in the OpenAI router accepts any authenticated user and forwards requests directly to upstream LLM providers without enforcing per-model access control. While the primary chat completion endpoint generatechatcompletion checks model ownership, group membership, and...