Hard to Read, Easy to Jailbreak: How Visual Degradation Bypasses MLLM Safety Alignment

Recent advancements in visual context compression enable MLLMs to process ultra-long contexts efficiently by rendering text into images. However, we identify a critical vulnerability inherent to this paradigm: lowering image resolution inadvertently catalyzes jailbreaking. Our experiments reveal...

PT-2026-37852

The issue was addressed with improved memory handling. This issue is fixed in Safari 26.3, iOS 26.3 and iPadOS 26.3, macOS Tahoe 26.3, visionOS 26.3. Processing maliciously crafted web content may lead to an unexpected process crash...

PT-2026-38061

The issue was addressed with improved memory handling. This issue is fixed in Safari 26.3, iOS 18.7.5 and iPadOS 18.7.5, iOS 26.3 and iPadOS 26.3, macOS Tahoe 26.3, visionOS 26.3. A remote attacker may be able to cause a denial-of-service...

Langchain-Chatchat Uses a Broken or Risky Cryptographic Algorithm

A flaw has been found in chatchat-space Langchain-Chatchat up to 0.3.1.3. This issue affects the function PIL.Image.tobytes of the file libs/chatchat-server/chatchat/webuipages/dialogue/dialogue.py of the component Vision Chat Paste Image Handler. This manipulation of the argument...

CVE-2026-7845

A flaw has been found in chatchat-space Langchain-Chatchat up to 0.3.1.3. This issue affects the function PIL.Image.tobytes of the file libs/chatchat-server/chatchat/webuipages/dialogue/dialogue.py of the component Vision Chat Paste Image Handler. This manipulation of the argument...

LangChain-Chatchat 加密问题漏洞

LangChain-Chatchat is a chatbot software developed based on the LangChain framework, open-sourced by Chatchat-Space. Versions of Langchain-Chatchat 0.3.1.3 and earlier had encryption-related vulnerabilities. These vulnerabilities stemmed from an issue in the Vision Chat Paste Image Handler...

PT-2026-37082

Name of the Vulnerable Software and Affected Versions Langchain-Chatchat versions prior to 0.3.1.4 Description A flaw in the Vision Chat Paste Image Handler component allows for the use of a weak hash. This occurs during the manipulation of the paste image.image data argument within the...

CVE-2026-42367

A privilege escalation vulnerability exists in the Web Interface / ssi.cgi functionality of GeoVision LPC2011/LPC2211 1.10. A specially crafted HTTP request can lead to credentials leak. An attacker can visit a webpage to trigger this vulnerability...

EUVD-2026-26857

Multiple reflected cross-site scripting xss vulnerabilities exist in the Web Interface / ssi.cgi functionality of GeoVision LPC2011/LPC2211 1.10. A specially crafted malicious url can lead to an arbitrary javascript code execution. An attacker can provide a crafted URL to trigger this vulnerabili...

CVE-2026-7161

An insufficient encryption vulnerability exists in the Device Authentication functionality of GeoVision GV-IP Device Utility 9.0.5. Listening to broadcast packets can lead to credentials leak. An attacker can listen to broadcast messages to trigger this vulnerability. When interacting with variou...

STARE: Step-Wise Temporal Alignment and Red-Teaming Engine for Multi-Modal Toxicity Attack

Red-teaming Vision-Language Models is essential for identifying vulnerabilities where adversarial image-text inputs trigger toxic outputs. Existing approaches treat image generation as a black box, returning only terminal toxicity scores and leaving open the question of when and how toxic semanti...

K000161051: WebKitGTK vulnerability CVE-2026-20652

Security Advisory Description The issue was addressed with improved memory handling. This issue is fixed in Safari 26.3, iOS 18.7.5 and iPadOS 18.7.5, iOS 26.3 and iPadOS 26.3, macOS Tahoe 26.3, visionOS 26.3. A remote attacker may be able to cause a denial-of-service. CVE-2026-20652 Impact There...

Hanwha Vision QND-8080R 操作系统命令注入漏洞

Hanwha Vision QND-8080R is a network infrared surveillance camera device produced by Hanwha Vision in South Korea. The Hanwha Vision QND-8080R has a vulnerability related to operating system command injection. This vulnerability arises from insufficient input validation, which may allow specially...

Hanwha Vision QND-8080R 输入验证错误漏洞

Hanwha Vision QND-8080R is a network infrared surveillance camera device produced by Hanwha Vision in South Korea. The Hanwha Vision QND-8080R has a vulnerability related to input validation errors. This vulnerability arises from improper handling of data in specific requests, which may lead to...

Hanwha Vision QND-8080R 访问控制错误漏洞

Hanwha Vision QND-8080R is a network infrared surveillance camera device produced by Hanwha Vision in South Korea. The Hanwha Vision QND-8080R has a access control vulnerability, which stems from issues with request processing. This vulnerability may lead to accidental access to protected functio...

PT-2026-35277

Name of the Vulnerable Software and Affected Versions GeoVision GV-IP Device Utility version 9.0.5 Description Insufficient encryption in the Device Authentication functionality allows for the leakage of administrator credentials. When the utility sends privileged commands to devices over UDP...

LMDeploy CVE-2026-33626 Flaw Exploited Within 13 Hours of Disclosure

A high-severity security flaw in LMDeploy, an open-source toolkit for compressing, deploying, and serving large language models LLMs, has come under active exploitation in the wild less than 13 hours after its public disclosure. The vulnerability, tracked as CVE-2026-33626 CVSS score: 7.5, relate...

CVE-2026-33626

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 LMDeploy's vision-language module. The loadimage function in lmdeploy/vl/utils.py fetches arbitrary URLs without validating...

VulnCheck KEV: CVE-2026-33626

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 LMDeploy's vision-language module. The loadimage function in lmdeploy/vl/utils.py fetches arbitrary URLs without validating...

Server-side Request Forgery (SSRF)

Overview Affected versions of this package are vulnerable to Server-side Request Forgery SSRF via the loadimage and encodeimagebase64 functions in LMDeploy's vision-language module, which fetch URLs without validating whether the destination is an internal or private address. An attacker can acce...