Originally reported by The Hacker News, MSRC Security Updates
TL;DR
Security researchers disclosed a high-severity vulnerability in OpenClaw that allows malicious websites to hijack local AI agents via WebSocket connections. Microsoft's security update guide also published three new CVEs affecting Vim and Vitess.
The OpenClaw vulnerability allows complete takeover of AI agents through malicious websites, representing a significant threat to users of AI automation tools with potential for widespread exploitation.
This week brought disclosure of vulnerabilities spanning AI automation tools, text editors, and database systems. The most significant finding involves a WebSocket-based attack against OpenClaw AI agents that requires no user interaction beyond visiting a malicious website.
Security researchers at Oasis disclosed a high-severity vulnerability in OpenClaw that enables malicious websites to connect to and control locally running AI agents. The flaw, dubbed "ClawJacked," affects the core OpenClaw gateway system without requiring any plugins or extensions.
According to the researchers, the vulnerability exists in the base OpenClaw installation running with default configuration. When successfully exploited, an attacker-controlled website can establish a WebSocket connection to the local AI agent and execute arbitrary commands or access sensitive data.
OpenClaw has released a fix for this issue. Organizations using OpenClaw for AI automation should update immediately and review their network security configurations to limit unnecessary exposure of local services.
Microsoft published details for CVE-2026-28417, an OS command injection vulnerability in Vim's netrw plugin. The netrw plugin provides file browsing and remote editing capabilities within Vim, making it a commonly used component.
Command injection vulnerabilities in text editors can be particularly dangerous as they may allow attackers to execute arbitrary system commands when users open specially crafted files or navigate to malicious network locations through the editor interface.
Vim users should update to the latest version and exercise caution when opening files from untrusted sources or browsing remote directories.
A second Vim vulnerability, CVE-2026-28422, involves a stack buffer overflow in the build_stl_str_hl() function. Stack buffer overflows can potentially lead to code execution or denial of service conditions depending on exploitation techniques and system protections.
This vulnerability likely affects Vim's status line processing functionality. The combination of two Vim vulnerabilities in the same disclosure cycle suggests organizations should prioritize updating their Vim installations across development and administrative systems.
CVE-2026-27969 affects Vitess, a database clustering system for horizontal scaling of MySQL. The vulnerability allows users with backup storage access to write files to arbitrary paths during database restoration operations.
Path traversal vulnerabilities in backup systems pose significant risks as they can enable attackers to overwrite critical system files or place malicious content in sensitive locations. Organizations using Vitess for database clustering should review backup access controls and update to the patched version.
The vulnerability requires existing backup storage access, limiting the attack surface to users with legitimate backup permissions who may have been compromised or malicious insiders.
Originally reported by The Hacker News, MSRC Security Updates