ai-captcha-bypass

AI-Powered CAPTCHA Solver This project is a Python-based command-line tool that uses large multimodal models LMMs like OpenAI's GPT-4o and Google's Gemini to automatically solve various types of CAPTCHAs. It leverages Selenium for web browser automation to interact with web pages and solve CAPTCH...

Which Image Property Carries the Jailbreak? A Controlled Dissection of Image-To-Text Jailbreaks

Image-to-text jailbreaks place harmful intent in text, image content, or the relationship between them. We examine image-side factors across four published attack families on a 313-prompt StrongREJECT slice, using five multimodal models and an additional appendix evaluation of InternVL3.5-8B. The...

The Poisoned Conversation: Privacy-Leaking Watermarks in Unified Multimodal Models

Multimodal models are increasingly shifting toward unified architectures that understand and generate text, images, and other modalities within a shared conversational context. This design enables fluid interaction across modalities, but it also changes the privacy threat model: Information...

Unveiling Privacy Risks in Multi-Modal Large Language Models: Task-Specific Vulnerabilities and Mitigation Challenges

Privacy risks in text-only Large Language Models LLMs are well studied, particularly their tendency to memorize and leak sensitive information. However, Multi-modal Large Language Models MLLMs, which process both text and images, introduce unique privacy challenges that remain underexplored...

Jailbreaking Frontier Foundation Models through Intention Deception

Large vision-language models exhibit remarkable capability but remain highly susceptible to jailbreaking. Existing safety training approaches aim to have the model learn a refusal boundary between safe and unsafe, based on the user's intent. It has been found that this binary training regime ofte...

Stealthy and Adjustable Text-Guided Backdoor Attacks on Multimodal Pretrained Models

Multimodal pretrained models are vulnerable to backdoor attacks, yet most existing methods rely on visual or multimodal triggers, which are impractical since visually embedded triggers rarely occur in real-world data. To overcome this limitation, we propose a novel Text-Guided Backdoor TGB attack...

CVE-2026-22773

A flaw was found in vLLM, an inference and serving engine for large language models LLMs. A remote attacker can exploit this vulnerability by sending a specially crafted 1x1 pixel image to a vLLM engine serving multimodal models that use the Idefics3 vision model implementation. This leads to a...

PYSEC-2026-143

vLLM is an inference and serving engine for large language models LLMs. In versions from 0.6.4 to before 0.12.0, users can crash the vLLM engine serving multimodal models that use the Idefics3 vision model implementation by sending a specially crafted 1x1 pixel image. This causes a tensor dimensi...

PYSEC-2026-143

vLLM is an inference and serving engine for large language models LLMs. In versions from 0.6.4 to before 0.12.0, users can crash the vLLM engine serving multimodal models that use the Idefics3 vision model implementation by sending a specially crafted 1x1 pixel image. This causes a tensor dimensi...

CVE-2026-22773

vLLM is an inference and serving engine for large language models LLMs. In versions from 0.6.4 to before 0.12.0, users can crash the vLLM engine serving multimodal models that use the Idefics3 vision model implementation by sending a specially crafted 1x1 pixel image. This causes a tensor dimensi...

CVE-2026-22773 vLLM is vulnerable to DoS in Idefics3 vision models via image payload with ambiguous dimensions

vLLM is an inference and serving engine for large language models LLMs. In versions from 0.6.4 to before 0.12.0, users can crash the vLLM engine serving multimodal models that use the Idefics3 vision model implementation by sending a specially crafted 1x1 pixel image. This causes a tensor dimensi...

CVE-2026-22773

CVE-2026-22773 affects vLLM (inference/serving engine) versions 0.6.4 through before 0.12.0 that serve multimodal models using the Idefics3 vision model. A crafted 1x1 pixel image triggers a tensor dimension mismatch in the image input processing, causing an unhandled runtime error and enabling a...

CVE-2026-22773 vLLM is vulnerable to DoS in Idefics3 vision models via image payload with ambiguous dimensions

vLLM is an inference and serving engine for large language models LLMs. In versions from 0.6.4 to before 0.12.0, users can crash the vLLM engine serving multimodal models that use the Idefics3 vision model implementation by sending a specially crafted 1x1 pixel image. This causes a tensor dimensi...

PT-2026-2260

Name of the Vulnerable Software and Affected Versions vLLM versions 0.6.4 through 0.11.9 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 that utilize the Idefics3 vision model implementatio...

CVE-2026-22773: Allocation of Resources Without Limits or Throttling

vLLM is an inference and serving engine for large language models LLMs. In versions from 0.6.4 to before 0.12.0, users can crash the vLLM engine serving multimodal models that use the Idefics3 vision model implementation by sending a specially crafted 1x1 pixel image. This causes a tensor dimensi...

OpenRT: An Open-Source Red Teaming Framework for Multimodal LLMs

The rapid integration of Multimodal Large Language Models MLLMs into critical applications is increasingly hindered by persistent safety vulnerabilities. However, existing red-teaming benchmarks are often fragmented, limited to single-turn text interactions, and lack the scalability required for...

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...

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...

DualTAP: A Dual-Task Adversarial Protector for Mobile MLLM Agents

The reliance of mobile GUI agents on Multimodal Large Language Models MLLMs introduces a severe privacy vulnerability: screenshots containing Personally Identifiable Information PII are often sent to untrusted, third-party routers. These routers can exploit their own MLLMs to mine this data,...

Enhanced MLLM Black-Box Jailbreaking Attacks and Defenses

Multimodal large language models MLLMs comprise of both visual and textual modalities to process vision language tasks. However, MLLMs are vulnerable to security-related issues, such as jailbreak attacks that alter the model's input to induce unauthorized or harmful responses. The incorporation o...