LMDeploy - Server-Side Request Forgery

LMDeploy is a toolkit for compressing, deploying, and serving large language models. Versions prior to 0.12.3 have a Server-Side Request Forgery SSRF vulnerability in the vision-language module. The loadimage function in lmdeploy/vl/utils.py fetches arbitrary URLs without validating internal or...

CVE-2026-33626-Lab

CVE-2026-33626 — LMDeploy Vision-Language SSRF Lab Overview This repository reproduces CVE-2026-33626 , a Server-Side Request Forgery SSRF vulnerability in LMDeploy's vision-language image loading path. The vulnerable behavior occurs when LMDeploy receives an image URL and the server-side image...

Learning-to-Detect

Learning to Detect Unknown Jailbreak Attacks in Large Vision-Language Models Official implementation of “Learning to Detect Unknown Jailbreak Attacks in Large Vision-Language Models.” This repository contains the data-processing, hidden-state extraction, classifier training, safety-pattern...

Can Quantum Layers Offer Performance and Security Advantages in Vision-Language Models?

As quantum computing moves from the noisy intermediate-scale quantum NISQ era toward early fault-tolerant quantum computing EFTQC, quantum-enhanced AI is shifting from a conceptual idea to experimental research. Quantum components are expected to extend the capabilities of existing AI systems, ye...

Training AIs to Solve CAPTCHAs

Interesting research: Abstract : Vision-language models VLMs can solve visual CAPTCHAs without task-specific training, but the agents built on them approach every challenge from scratch. For such an agent, the hundredth instance of a familiar puzzle costs as much time and compute as the first...

DarKnight: Retrieval-Guided Agentic Framework for Breaking Dark Web CAPTCHA Gates

Cyber Threat Intelligence CTI investigators who monitor dark web marketplaces for leaked credentials and breach listings must pass the diverse CAPTCHAs that operators deploy to block automated collection. Existing solvers either need unscalable per-site fine-tuning or rely on commercial...

GeoCloak: An Agentic Framework for Semantics-Preserving Geolocation Privacy Via Peripheral Clue Manipulation

The rapid advancement of large vision language models LVLMs has made geolocation inference from images highly accurate and accessible. This poses serious privacy risks: a passive adversary with black-box LVLM access can infer a user's location from a single photo with no metadata. Existing...

Unmasking the Secret Code: Using Canaries to Show How Masked Context Controls Memorization in SFT

Canaries, or synthetic training records containing known secrets, are widely used to assess whether language models reveal information from their training data. But how much do these measurements depend on the canaries themselves? We study this question through controlled fine-tuning experiments...

Understanding and Mitigating Token-Pruning-Induced Vulnerabilities in VLMs

Token-Pruning accelerates Vision-Language Models by removing redundant visual tokens, yet its safety implications remain underexplored. In this work, we present the first comprehensive safety evaluation of Token-Pruning mechanisms and find that: most pruning strategies significantly degrade safet...

Adversarial Images Hijack Web Agents from Visual Grounding to Browser Execution

Modern web agents built on large vision-language models process webpages, select relevant UI elements, and translate model outputs into browser actions. Existing visual red-teaming approaches use adversarial visual content to manipulate this process. However, they primarily target model inference...

Exploiting Vulnerabilities: Universal Adversarial Attacks on Vision-Language-Action Models in Robotics

Recently, Vision-Language-Action VLA models have revolutionized robotic manipulation by seamlessly integrating visual perception, language understanding, and action generation in an end-to-end learning framework. However, since these models are designed to interact directly with the physical worl...

Selective Channel Restoration for Backdoored Vision-Language Models

Vision-language models VLMs exhibit strong multimodal capabilities but remain vulnerable to backdoors implanted through poisoned fine-tuning data. Existing defenses often require extensive parameter updates during fine-tuning or incur per-query overhead during inference. To address these...

Backdoors in Learning-Based Industrial Robotic Arm Manipulation: An Empirical Security Study

Learning-based models e.g., visuomotor and Vision-Language-Action VLA are increasingly explored for industrial robotic manipulation, where model predictions are directly translated into physical actions. This tight coupling between model behavior and physical execution makes hidden security...

Repeat-After-Me: Black-Box Adaptive Visual Prompt Injection

Prompt injection is widely recognized as a major security threat to AI agents that interact with untrusted external data, such as websites, documents, and emails. Prior work has shown that, in the text domain, black-box prompt injection can achieve near-perfect attack success rates ASRs. In the...

CAPTCHAs in the Agentic Era: Solvers That Learn from Every Encounter

Vision-language models VLMs can solve visual CAPTCHAs without task-specific training, but the agents built on them approach every challenge from scratch. For such an agent, the hundredth instance of a familiar puzzle costs as much time and compute as the first. Specialized detectors invert the...

Bit-Flip Attacks on Vision-Language-Action Models: Action-Decoding Architecture Shapes the Vulnerability

Quantized Vision-Language-Action VLA models expose a weight-fault surface: Rowhammer-style faults can corrupt deployed INT8 bits. We present the first bit-flip attack on a VLA: a few gradient-selected flips reduce closed-loop success to $0%$, while hundreds of random flips are harmless. Across...

Once Poisoned, Arbitrarily Controlled: A Programmable Backdoor in VLMs

Existing vision-language model VLM backdoors are usually treated as static vulnerabilities: one-to-one and N-to-N attacks bind one or more triggers to a finite set of targets before victim training. This assumption substantially underestimates the threat. We show that a single poisoning phase can...

Decoy Images Amplify Caption-Mediated Defenses against Encoded Jailbreaks

We report a counter-intuitive interaction between image inputs and existing black-box defenses on Vision--Language Models VLMs: pairing an encoded jailbreak prompt with an unrelated decoy image can sharply lower attack success rate ASR. The operative change is in the defense pipeline, not in the...

MulRobBench: A Decision-Level Benchmark for Safe and Security-Policy-Compliant Multimodal UAV Agents

Smart-city airspace is transforming Uncrewed Aerial Vehicles UAVs from passive sensing platforms into cyber-physical decision makers that must follow operational rules under degraded observations and ambiguous language. Existing UAV and multimodal benchmarks evaluate perception, navigation,...

PYSEC-2026-2607 LMDeploy has Server-Side Request Forgery (SSRF) via Vision-Language Image Loading

Summary A Server-Side Request Forgery SSRF vulnerability exists in LMDeploy's vision-language module. The loadimage function in lmdeploy/vl/utils.py fetches arbitrary URLs without validating internal/private IP addresses, allowing attackers to access cloud metadata services, internal networks, an...