NEWS

Unpatched Tailscale SSH Flaws Could Let a Hacked Device Reach Root on Other Servers

The 'Tailworm' report says Tailscale SSH port forwarding enables root escalation from one compromised device. No CVE or fix yet.

Dylan H.

News Desk

October 26, 2026
3 min read
Unpatched Tailscale SSH Flaws Could Let a Hacked Device Reach Root on Other Servers

Overview

Researcher Erica Windisch of Cyberdione Labs published a report called "Tailworm" on October 25, describing how a single compromised device on a Tailscale network can be used to gain root on other servers it can reach. The testing was done in a lab against Tailscale 1.104.2 and headscale 0.29.5, not against live networks.

At publication there was no CVE, security bulletin, or fixed version, and Tailscale had not confirmed the specific technical claims.

Attack Conditions

The attack needs:

  1. Tailscale SSH configured to allow port forwarding
  2. An attacker who already controls a device that access rules allow to reach the target

No stolen passwords or keys are required.

Escalation Paths

The report lists 11 findings overall. Three escalation methods stand out:

  • Standard port forwarding caused a root-level process to access a root-only socket, exposing the system password file
  • Socket replacement of a root-owned socket allowed direct root command execution against certain services
  • A race condition in a file-access check let the attacker capture a device-enrollment key and enroll their own machine as a trusted server. This last step only worked where Tailnet Lock was disabled

Windisch calls the threat "wormable," though some researchers dispute that. Single source-to-destination rules and hardware-bound device keys limit spread.

Vendor Response

Tailscale VP of engineering Mike Shaver said the company has received the report, has found issues it is working to fix, and is still determining which setups are affected. Tailscale does not recommend disabling Tailscale SSH entirely. Some commenters criticized the disclosure as uncoordinated; Windisch says she sent the report and proof-of-concept code to Tailscale's security contact before publishing.

Mitigations

  • Write access rules as single device-to-target pairs and remove broad identity or tag grants
  • Disable SSH forwarding where possible. On headscale 0.29.5 it cannot be disabled, so treat every SSH grant as root access
  • Remove unneeded app connector permissions and any serve forwards pointing at Unix sockets
  • Enable node-key expiry, enable Tailnet Lock, and delete compromised devices at the control plane
  • SSH check mode and automatic updates add friction but do not fix the escalation

Takeaway

Flat overlay networks turn one compromised endpoint into a stepping stone. Until a fix ships, tighten ACLs and audit which devices hold port-forwarding SSH grants.

Sources