CVE-2026-53811

Vulnerability

OpenClaw before version 2026.5.7 contains a privilege escalation vulnerability in the allowFrom feature for Matrix identities. The feature binds policy entries to mutable display name metadata rather than stable Matrix user IDs. This allows an authenticated account that can change its display name to match a policy entry intended for another Matrix identity [1][2].

Exploitation

An attacker must have an authenticated Matrix account with the ability to change their display name. By altering their display name to match the display name of a target identity that has been granted agent access via the allowFrom feature, the attacker can receive agent access intended for that target. No additional privileges or user interaction are required beyond the ability to change display names [1][2].

Impact

Successful exploitation allows the attacker to gain unauthorized agent access, potentially leading to privilege escalation within the OpenClaw gateway. The practical impact depends on the operator's configuration and whether lower-trust input can reach the affected path. This could result in unauthorized actions, data access, or further compromise depending on the permissions granted to the agent [2].

Mitigation

The first stable patched version is 2026.5.7 [2]. Operators should upgrade to this version or later. As a workaround, use stable Matrix user IDs in allowlists instead of display names, keep channel and tool allowlists narrow, avoid sharing a gateway between mutually untrusted users, and disable the affected feature when not needed [2]. No KEV listing is currently available.