CVE-2026-0599 Unbounded External Image Fetch in Validation Leads to Resource-Exhaustion DoS in huggingface/text-generation-inference

A vulnerability in huggingface/text-generation-inference version 3.3.6 allows unauthenticated remote attackers to exploit unbounded external image fetching during input validation in VLM mode. The issue arises when the router scans inputs for Markdown image links and performs a blocking HTTP GET...

CVE-2026-0599 Unbounded External Image Fetch in Validation Leads to Resource-Exhaustion DoS in huggingface/text-generation-inference

A vulnerability in huggingface/text-generation-inference version 3.3.6 allows unauthenticated remote attackers to exploit unbounded external image fetching during input validation in VLM mode. The issue arises when the router scans inputs for Markdown image links and performs a blocking HTTP GET...

CVE-2026-0599 Unbounded External Image Fetch in Validation Leads to Resource-Exhaustion DoS in huggingface/text-generation-inference

A vulnerability in huggingface/text-generation-inference version 3.3.6 allows unauthenticated remote attackers to exploit unbounded external image fetching during input validation in VLM mode. The issue arises when the router scans inputs for Markdown image links and performs a blocking HTTP GET...

DF-LoGiT: Data-Free Logic-Gated Backdoor Attacks in Vision Transformers

The widespread adoption of Vision Transformers ViTs elevates supply-chain risk on third-party model hubs, where an adversary can implant backdoors into released checkpoints. Existing ViT backdoor attacks largely rely on poisoned-data training, while prior data-free attempts typically require...

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...

Embracing Choice in Cybersecurity: TrendAI Vision One™ and SentinelOne Integration

Discover how the TrendAI Vision One and SentinelOne integration exemplifies our commitment to endpoint flexibility...

CVE-2021-47795

GeoVision GeoWebServer 5.3.3 contains multiple vulnerabilities including local file inclusion, cross-site scripting, and remote code execution through improper input sanitization. Attackers can exploit the WebStrings.srf endpoint by manipulating path traversal and injection parameters to access...

Hanwha Vision Camera Use of Hard-coded Cryptographic Key (CVE-2025-52601)

Cybersecurity Nozomi Networks Labs, a specialized security company focused on Industrial Control Systems ICS and OT/IoT security, has discovered a vulnerability in Device Manager that a hardcoded encryption key for sensitive information. An attacker can use key to decrypt sensitive information. T...

Hanwha Vision Camera Improper Privilege Management (CVE-2025-52599)

Cybersecurity Nozomi Networks Labs, a specialized security company focused on Industrial Control Systems ICS and OT/IoT security, has discovered Inadequate of permission management for camera guest account. The manufacturer has released patch firmware for the flaw, please refer to the...

Hanwha Vision Camera Improper Neutralization of Input During Web Page Generation (CVE-2025-8075)

Cybersecurity Nozomi Networks Labs, a specialized security company focused on Industrial Control Systems ICS and OT/IoT security, has discovered that validation of incoming XML format request messages is inadequate. This vulnerability could allow an attacker to XSS on the user's browser. The...

Hanwha Vision Camera Improper Certificate Validation(CVE-2025-52598)

Cybersecurity Nozomi Networks Labs, a specialized security company focused on Industrial Control Systems ICS and OT/IoT security, has found a flaw that camera's client service does not perform certificate validation. The manufacturer has released patch firmware for the flaw, please refer to the...

Hanwha Vision Camera Improper Input Validation (CVE-2025-52600)

Cybersecurity Nozomi Networks Labs, a specialized security company focused on Industrial Control Systems ICS and OT/IoT security, has discovered a vulnerability in camera video analytics that Improper input validation. This vulnerability could allow an attacker to execute specific commands on the...

CVE-2021-47795

GeoVision GeoWebServer 5.3.3 contains multiple vulnerabilities including local file inclusion, cross-site scripting, and remote code execution through improper input sanitization. Attackers can exploit the WebStrings.srf endpoint by manipulating path traversal and injection parameters to access...

Unity Linux 20.1060a / 20.1070a Security Update: kernel (UTSA-2026-002277)

The Unity Linux 20 host has a package installed that is affected by a vulnerability as referenced in the UTSA-2026-002277 advisory. The usbvision driver in the Linux kernel package 3.10.0-123.20.1.el7 through 3.10.0-229.14.1.el7 in Red Hat Enterprise Linux RHEL 7.1 allows physically proximate...

GHSA-GRG2-63FW-F2QR vLLM is vulnerable to DoS in Idefics3 vision models via image payload with ambiguous dimensions

Summary 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 dimension mismatch that results in an unhandled runtime error, leading to complete server termination. Details T...

Key Insights on SHADOW-AETHER-015 and Earth Preta from the 2025 MITRE ATT&CK Evaluation with Trend Vision One™

This blog discusses notable modern TTPs observed from SHADOW-AETHER-015 and Earth Preta, from Trend Research™ monitoring and Trend Vision One™ intelligence. These findings support the performance of TrendAI™ in the 2025 MITRE ATT&CK Evaluations...

Key Insights on SHADOW-AETHER-015 and Earth Preta from the 2025 MITRE ATT&CK Evaluation with TrendAI Vision One™

This blog discusses notable modern TTPs observed from SHADOW-AETHER-015 and Earth Preta, from TrendAI Research™ monitoring and TrendAI Vision One™ intelligence. These findings support the performance of TrendAI™ in the 2025 MITRE ATT&CK Evaluations...

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 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...