AI Sandboxes: A Threat Model, Taxonomy, and Measurement Framework

AI systems are increasingly evaluated in bounded environments that combine isolation, simulation, instrumentation, supervision, and evidence capture. For physical AI, AIoT, and cyber-physical systems, this shift is not a matter of terminology: the system under test may sense, decide, actuate,...

PT-2026-57550

Name of the Vulnerable Software and Affected Versions Crawl4AI versions prior to 0.8.8 Description The Docker API server contains credential exfiltration flaws due to improper input validation and insecure configuration handling. Unauthenticated attackers can redirect LLM API calls to external...

Malicious Package

Overview web-model-bridge is a malicious package. This package contains malicious code, and its content was removed from the official package manager. While this package might be attempting to impersonate a valid organization, there is no connection between that organization and this package...

MINI-CGMC-V394-687H

Bulletin has no description...

UAParser.js: Unbounded `Sec-CH-UA-Model` parsing can trigger ReDoS in `withClientHints()`

Summary A regular expression denial-of-service ReDoS vulnerability has been discovered in ua-parser-js when using the Client Hints API. By sending a crafted Sec-CH-UA-Model header to an application that calls UAParserheaders.withClientHints, an attacker can cause the parser to spend excessive CPU...

Regular Expression Denial of Service (ReDoS)

Overview ua-parser-js is a lightweight JavaScript-based user-agent string parser. Affected versions of this package are vulnerable to Regular Expression Denial of Service ReDoS via the withClientHints function. An attacker can cause excessive CPU consumption by sending a specially crafted...

NPM: UAParser.js: Unbounded `Sec-CH-UA-Model` parsing can trigger ReDoS in `withClientHints()`

NPM: UAParser.js: Unbounded Sec-CH-UA-Model parsing can trigger ReDoS in withClientHints vulnerability discovered by ? in WordPress Npm ua-parser-js versions = 2.0.1, 2.0.10...

MINI-MVG6-456M-P46R

Bulletin has no description...

MINI-M3MQ-HM99-6P3V

Bulletin has no description...

LiteLLM Vulnerability Chain Lets Low-Privilege Users Take Over AI Gateway Servers

A default low-privilege account on a LiteLLM proxy can climb to full admin and run code on the server by chaining three vulnerabilities, researchers at Obsidian Security disclosed LiteLLM is a widely deployed open-source AI gateway that brokers calls to more than 100 model providers behind one...

MINI-P644-HCCC-F9WH

Bulletin has no description...

Security Bulletin: IBM Engineering Systems Design Rhapsody TestConductor was affected by CVE-2025-48924

Summary IBM Engineering Systems Design Rhapsody TestConductor was vulnerable to an uncontrolled recursion on very long inputs. This could cause components using Apache Commons Lang to stop. Vulnerability Details CVEID:CVE-2025-48924 DESCRIPTION: Uncontrolled Recursion vulnerability in Apache...

CVE-2026-34022

The CVE-2026-34022 entry affects Wertheim SafeController Family 65000, Controller 65000 (AssemblyVersion 6.11.8130.22319). The root cause is the use of weak custom cryptographic algorithms with hard-coded keys to protect communications, enabling interception of data in transit. During reassessmen...

CVE-2026-45833

A flaw was found in the ChromaDB Python project. An authenticated attacker with UPDATECOLLECTION permission could exploit a code injection vulnerability. By sending a malicious model repository to a specific API endpoint with trustremotecode enabled, the attacker can execute arbitrary code on the...

CVE-2026-12223

A vulnerability was identified in Yealink SIP-T46U 108.86.0.118. Affected by this vulnerability is the function modwebd.TFTPUploadIperf of the file /api/inner/tftpuploadiperf of the component Web FastCGI Service. The manipulation of the argument ip/port leads to command injection. The attack need...

CVE-2026-12220

A vulnerability exists in Yealink SIP-T46U firmware 108.86.0.118 affecting the mod_upgrade.SparePartsUpload handler in /api/upgrade/accupgradebychunk. Manipulating the uid argument can cause a stack-based buffer overflow. Exploitation is described as local-network only, with public disclosure and...

How Much Can We Trust LLM Search Agents? Measuring Endorsement Vulnerability to Web Content Manipulation

Large language model LLM-based search agents synthesize open-web content into actionable recommendations on behalf of users, creating a risk that attacker-published pages are transformed into endorsed claims. We introduce SearchGEO, a controlled evaluation framework for measuring endorsement...

SoK: Security and Privacy of Foundation-Model-Powered Robots

Foundation models are reshaping robotics by enabling robots to interpret open-ended instructions, reason over multimodal contexts, and operate in complex, open-world environments. However, their integration also introduces security and privacy S&P risks that extend beyond the FMs themselves to...

CVE-2026-38062

The Tenda 5G03 router, specifically version V05.03.02.04 (Version 1.0) , is vulnerable to a Command injection flaw. The vulnerability exists in the action_set_rat_mode() function, where the ratMode parameter is improperly neutralized, allowing for the execution of arbitrary OS commands. This issu...

PT-2026-49551

Name of the Vulnerable Software and Affected Versions UAParser.js versions 2.0.1 through 2.0.9 Description A regular expression denial-of-service ReDoS exists when using the Client Hints API. An attacker can cause excessive CPU consumption and a denial-of-service condition in server-side...