vLLM 0.8.3 - 0.14.0 - Information Disclosure

vLLM 0.8.3 to - 0.14.1 contains an information disclosure caused by leaking a heap address in error messages from the multimodal endpoint when processing invalid images, letting remote attackers reduce ASLR entropy, exploit requires sending invalid images. id: CVE-2026-22778 info: name: vLLM 0.8....

bordair-multimodal

Dataset di Iniezione di Prompt Multimodale 516.588 campioni etichettati 251.782 attacco + 251.576 benigni, più una suddivisione di validazione reale di 13.230 campioni in cinque versioni del dataset più l'ingestione di dataset esterni, coprendo attacchi cross-modali, multi-turn, suffisso...

hcaptcha-challenger

hCaptcha Challenger 🚀 Enfrente o desafio hCaptcha com elegância usando um modelo de linguagem de grande porte multimodal. Introdução Não depende de nenhum script Tampermonkey. Não utiliza nenhum serviço anti-captcha de terceiros. Apenas implementa algumas interfaces para tornar possível o AI vs A...

Learning-to-Detect

Imparare a Rilevare Attacchi Jailbreak Sconosciuti nei Grandi Modelli Visione-Linguaggio Questo repository fornisce l'implementazione ufficiale di "Learning to Detect Unknown Jailbreak Attacks in Large Vision-Language Models". Contenuto Modello base Rilevamento degli Attacchi Jailbreak Dataset...

Multimodal-Unlearnable-Examples

मल्टीमॉडल अनलर्नेबल उदाहरण: मल्टीमॉडल कंट्रास्टिव लर्निंग के विरुद्ध डेटा की सुरक्षा ACM MM2024 सारांश : मल्टीमॉडल कंट्रास्टिव लर्निंग MCL ने इंटरनेट से क्रॉल किए गए लाखों इमेज-कैप्शन जोड़ों से सीखकर जीरो-शॉट वर्गीकरण में उल्लेखनीय प्रगति दिखाई है। हालांकि, यह निर्भरता गोपनीयता जोखिम पैदा करती है...

Multimodal-web-attack-Dataset

MWAD : un jeu de données multimodal d'attaques web pour la détection d'injections SQL par IA Vue d'ensemble Le jeu de données multimodal d'attaques web MWAD est un jeu de données de cybersécurité inédit et étiqueté, développé dans le cadre des recherches de l'auteur en cybersécurité à l'Universit...

ContextHound

ContextHound Statisches Analysetool, das Ihre Codebasis auf LLM-Prompt-Injection- und multimodale Sicherheitslücken scannt. Läuft offline, keine API-Aufrufe erforderlich. Das ContextHound-Ökosystem ContextHound ist in Ihrem gesamten Entwicklungs- und Browser-Workflow verfügbar:...

ai-captcha-bypass

Risolutore di CAPTCHA basato su IA Questo progetto è uno strumento da riga di comando basato su Python che utilizza grandi modelli multimodali LMM come GPT-4o di OpenAI e Gemini di Google per risolvere automaticamente vari tipi di CAPTCHA. Sfrutta Selenium per l'automazione del browser web al fin...

CVE-2026-47617

NVIDIA Dynamo for Linux contains a vulnerability in the multimodal media fetcher where an attacker may cause server-side request forgery via DNS rebinding. A successful exploit of this vulnerability might lead to information disclosure...

CVE-2026-47616

NVIDIA Dynamo for Linux contains a vulnerability in the multimodal media fetcher where an attacker may cause server-side request forgery. A successful exploit of this vulnerability might lead to information disclosure...

CVE-2026-47618

NVIDIA Dynamo for Linux contains a vulnerability in the Rust multimodal media fetcher where an attacker could cause server-side request forgery. A successful exploit of this vulnerability might lead to information disclosure...

CVE-2026-47615

NVIDIA Dynamo for Linux contains a vulnerability where an attacker may cause server-side request forgery by supplying a crafted URL in a multimodal request. A successful exploit of this vulnerability might lead to information disclosure...

CVE-2026-47613

NVIDIA Dynamo for Linux contains a vulnerability where an attacker may cause improper limitation of a pathname to a restricted directory by supplying a crafted local path in a multimodal request. A successful exploit of this vulnerability might lead to information disclosure...

CVE-2026-24255

NVIDIA Dynamo for Linux contains a vulnerability in the multimodal embedding cache, where an attacker could cause a hash collision by submitting images that share an identical pixel byte sequence but have different dimensions. A successful exploit of this vulnerability might lead to data tamperin...

CVE-2026-24254

NVIDIA Dynamo for Linux contains a vulnerability in the multimodal serving topology, where an attacker could cause an out-of-bounds write. A successful exploit of this vulnerability might lead to code execution, escalation of privileges, data tampering, denial of service, and information disclosu...

EUVD-2026-52813

NVIDIA Dynamo for Linux contains a vulnerability in the Rust multimodal media fetcher where an attacker could cause server-side request forgery. A successful exploit of this vulnerability might lead to information disclosure...

CVE-2026-47618

NVIDIA Dynamo for Linux contains a vulnerability in the Rust multimodal media fetcher where an attacker could cause server-side request forgery. A successful exploit of this vulnerability might lead to information disclosure...

CVE-2026-47618

NVIDIA Dynamo for Linux contains a vulnerability in the Rust multimodal media fetcher where an attacker could cause server-side request forgery. A successful exploit of this vulnerability might lead to information disclosure...

EUVD-2026-52812

NVIDIA Dynamo for Linux contains a vulnerability in the multimodal media fetcher where an attacker may cause server-side request forgery via DNS rebinding. A successful exploit of this vulnerability might lead to information disclosure...

CVE-2026-47617

NVIDIA Dynamo for Linux contains a vulnerability in the multimodal media fetcher where an attacker may cause server-side request forgery via DNS rebinding. A successful exploit of this vulnerability might lead to information disclosure...