Odysseus: Jailbreaking Commercial Multimodal LLM-Integrated Systems Via Dual Steganography

By integrating language understanding with perceptual modalities such as images, multimodal large language models MLLMs constitute a critical substrate for modern AI systems, particularly intelligent agents operating in open and interactive environments. However, their increasing accessibility al...

Chinese Surveillance and AI

New report: "The Party's AI: How China's New AI Systems are Reshaping Human Rights." From a summary article: China is already the world's largest exporter of AI powered surveillance technology; new surveillance technologies and platforms developed in China are also not likely to simply stay there...

Cisco Integrated AI Security and Safety Framework Report

Artificial intelligence AI systems are being readily and rapidly adopted, increasingly permeating critical domains: from consumer platforms and enterprise software to networked systems with embedded agents. While this has unlocked potential for human productivity gains, the attack surface has...

RunawayEvil: Jailbreaking the Image-To-Video Generative Models

Image-to-Video I2V generation synthesizes dynamic visual content from image and text inputs, providing significant creative control. However, the security of such multimodal systems, particularly their vulnerability to jailbreak attacks, remains critically underexplored. To bridge this gap, we...

AgenticCyber: A GenAI-Powered Multi-Agent System for Multimodal Threat Detection and Adaptive Response in Cybersecurity

The increasing complexity of cyber threats in distributed environments demands advanced frameworks for real-time detection and response across multimodal data streams. This paper introduces AgenticCyber, a generative AI powered multi-agent system that orchestrates specialized agents to monitor...

OmniSafeBench-MM: A Unified Benchmark and Toolbox for Multimodal Jailbreak Attack-Defense Evaluation

Recent advances in multi-modal large language models MLLMs have enabled unified perception-reasoning capabilities, yet these systems remain highly vulnerable to jailbreak attacks that bypass safety alignment and induce harmful behaviors. Existing benchmarks such as JailBreakV-28K, MM-SafetyBench,...

COGNITION: From Evaluation to Defense against Multimodal LLM CAPTCHA Solvers

This paper studies how multimodal large language models MLLMs undermine the security guarantees of visual CAPTCHA. We identify the attack surface where an adversary can cheaply automate CAPTCHA solving using off-the-shelf models. We evaluate 7 leading commercial and open-source MLLMs across 18...

CVE-2025-62372

vLLM is an inference and serving engine for large language models LLMs. From version 0.5.5 to before 0.11.1, users can crash the vLLM engine serving multimodal models by passing multimodal embedding inputs with correct ndim but incorrect shape e.g. hidden dimension is wrong, regardless of whether...

CVE-2025-62372

vLLM is an inference and serving engine for large language models LLMs. From version 0.5.5 to before 0.11.1, users can crash the vLLM engine serving multimodal models by passing multimodal embedding inputs with correct ndim but incorrect shape e.g. hidden dimension is wrong, regardless of whether...

CVE-2025-62372

CVE-2025-62372 affects vLLM (inference/serving engine). From version 0.5.5 up to before 0.11.1, passing multimodal embedding inputs with correct ndim but incorrect shape (e.g., wrong hidden dimension) can crash the engine when serving multimodal models, regardless of whether those inputs are supp...

CVE-2025-62372 vLLM vulnerable to DoS with incorrect shape of multimodal embedding inputs

vLLM is an inference and serving engine for large language models LLMs. From version 0.5.5 to before 0.11.1, users can crash the vLLM engine serving multimodal models by passing multimodal embedding inputs with correct ndim but incorrect shape e.g. hidden dimension is wrong, regardless of whether...

CVE-2025-62372 vLLM vulnerable to DoS with incorrect shape of multimodal embedding inputs

vLLM is an inference and serving engine for large language models LLMs. From version 0.5.5 to before 0.11.1, users can crash the vLLM engine serving multimodal models by passing multimodal embedding inputs with correct ndim but incorrect shape e.g. hidden dimension is wrong, regardless of whether...

EUVD-2025-198357

vLLM is an inference and serving engine for large language models LLMs. From version 0.5.5 to before 0.11.1, users can crash the vLLM engine serving multimodal models by passing multimodal embedding inputs with correct ndim but incorrect shape e.g. hidden dimension is wrong, regardless of whether...

CVE-2025-62372 vLLM vulnerable to DoS with incorrect shape of multimodal embedding inputs

vLLM is an inference and serving engine for large language models LLMs. From version 0.5.5 to before 0.11.1, users can crash the vLLM engine serving multimodal models by passing multimodal embedding inputs with correct ndim but incorrect shape e.g. hidden dimension is wrong, regardless of whether...

vLLM 输入验证错误漏洞

vLLM is a high throughput and memory efficient inference and service engine for LLM from the vLLM open source. An input validation error vulnerability exists in vLLM versions 0.5.5 through prior to 0.11.1, which stems from improper handling of multimodal embedded inputs and could cause the engine...

vLLM vulnerable to DoS with incorrect shape of multimodal embedding inputs

Summary Users can crash the vLLM engine serving multimodal models by passing multimodal embedding inputs with correct ndim but incorrect shape e.g. hidden dimension is wrong, regardless of whether the model is intended to support such inputs as defined in the Supported Models page. The issue has...

Improper Validation of Array Index

Overview vllm is an A high-throughput and memory-efficient inference and serving engine for LLMs Affected versions of this package are vulnerable to Improper Validation of Array Index via the MultiModalDataParser input processor. An attacker can cause the engine to crash by submitting multimodal...

GHSA-PMQF-X6X8-P7QW vLLM vulnerable to DoS with incorrect shape of multimodal embedding inputs

Summary Users can crash the vLLM engine serving multimodal models by passing multimodal embedding inputs with correct ndim but incorrect shape e.g. hidden dimension is wrong, regardless of whether the model is intended to support such inputs as defined in the Supported Models page. The issue has...

PT-2025-47649

Name of the Vulnerable Software and Affected Versions vLLM versions 0.5.5 through 0.11.0 Description vLLM is an inference and serving engine for large language models LLMs. Users can cause the vLLM engine to crash when serving multimodal models by providing multimodal embedding inputs with a...

Can MLLMs Detect Phishing? A Comprehensive Security Benchmark Suite Focusing on Dynamic Threats and Multimodal Evaluation in Academic Environments

The rapid proliferation of Multimodal Large Language Models MLLMs has introduced unprecedented security challenges, particularly in phishing detection within academic environments. Academic institutions and researchers are high-value targets, facing dynamic, multilingual, and context-dependent...