A researcher finds 11 unpatched root-access bugs in Tailscale SSH
Tailworm, a new report from Erica Windisch, shows one compromised device escalating to root on every other node a Tailscale network lets it reach, with no CVE or patch yet.
Security researcher Erica Windisch of Cyberdione Labs published Tailworm on 25 October, a report detailing 11 findings across Tailscale SSH, serve, Funnel, app connectors and Headscale, the open-source control server people self-host instead of Tailscale’s own. Tested against Tailscale 1.104.2 and Headscale 0.29.5, the worst finding is a Unix-socket handling bug in Tailscale SSH: because tailscaled runs as root, a forwarded connection opened with ssh -L or -R executes with root rights. Chained with the other findings, Windisch demonstrated one already-compromised device escalating to root on every other node its access rules allow, with no stolen password or key, in well under a second per hop.
No CVE has been issued and Tailscale has not shipped a fix. The company’s VP of engineering acknowledged the report on Reddit, saying Tailscale is already looking at the same areas with outside researchers but is “not recommending that everyone disable Tailscale SSH” while it works out which configurations are affected.
Why it matters: if your team uses Tailscale SSH or Headscale for remote access, the mitigations are things you can do today rather than wait on a patch for: restrict SSH access rules to specific device pairs instead of whole identities, and block TCP forwards to privileged Unix sockets.
The caveat: this was demonstrated in a lab, not observed as a live attack, and Windisch reported it to Tailscale without coordinating a disclosure date, so there is no official bulletin yet confirming full scope.