The developers of OpenClaw recently patched a critical vulnerability that could be exploited to hijack the increasingly popular AI assistant by tricking the target user into visiting a malicious website.
OpenClaw (previously known as Clawdbot and Moltbot) is an open source, self-hosted AI agent that can autonomously execute terminal commands, manage file systems, and orchestrate complex workflows across messaging apps.
Researchers at security firm DepthFirst discovered recently that OpenClaw is affected by a vulnerability that allows an attacker to obtain a user’s authentication token, which can be used to connect to the victim’s OpenClaw instance.
The security hole, tracked as CVE-2026-25253, was patched in recent days with the release of version 2026.1.29.
“This is a token exfiltration vulnerability that leads to full gateway compromise,” the AI tool’s developers explained in an advisory. “It impacts any Moltbot deployment where a user has authenticated to the Control UI. The attacker gains operator-level access to the gateway API, enabling arbitrary config changes and code execution on the gateway host.”
DepthFirst has shared a detailed technical explanation of how the attack works.
To initiate an attack, an attacker only needs to trick the victim into visiting a malicious website. The attacker’s site is designed to execute JavaScript in the victim’s browser to obtain an OpenClaw authentication token and send it back to the attacker.
The attacker’s site also executes code to establish a WebSocket connection to the local host, with authentication enabled using the stolen token. The attacker can then disable sandboxing, along with user confirmation for the execution of dangerous commands.
Because OpenClaw has elevated privileges on the system and deep access to data and applications, the attacker can gain access to valuable information belonging to the victim and abuse the AI assistant to execute arbitrary commands and take control of the host.
While the project is only a couple of months old, this is not the first time security issues have been highlighted.
In late January, Jamieson O’Reilly demonstrated that threat actors can exploit OpenClaw to access highly sensitive information, execute commands on the host system, and manipulate users.
Cisco’s security experts have also identified potentially severe issues and warned enterprises that OpenClaw can be a “security nightmare”.
Related: 175,000 Exposed Ollama Hosts Could Enable LLM Abuse
Related: LLMs Hijacked, Monetized in ‘Operation Bizarre Bazaar’
Related: Chrome, Edge Extensions Caught Stealing ChatGPT Sessions
