A researcher has identified flaws in Tailscale SSH that could let an attacker with a foothold on one device gain root access to other reachable servers, according to SC Media, reporting The Hacker News. The Tailworm report by Erica Windisch of Cyberdione Labs, published 25 October, was tested in a lab against Tailscale 1.104.2 and headscale 0.29.5 between 22 and 24 October.

What the report claims

The core finding is that Unix socket forwarding in Tailscale SSH runs as root, letting an admitted device reach root-only sockets, the report says. In the lab, local forwarding gave root access and allowed reading the password file, while remote forwarding replaced root-owned sockets and allowed root code execution. A timing gap could also let a rogue node join the network, and app connectors could reach cloud metadata and Kubernetes services.

The report lists 11 findings, including headscale issues where unauthenticated registrations could evict pending approvals. On headscale 0.29.5, forwarding cannot be disabled, the report says, while Tailscale's hosted service allows it per rule.

Lab findings, no fix yet

Three points matter. First, the attacks were lab demonstrations, not observed on real networks, and the proof of concept is not public. Second, they need a permitted attacker-controlled device plus forwarding enabled: they do not provide the initial foothold. Third, as of 26 October, no bulletin, CVE or fix had been identified, and Tailscale says it is investigating and working on fixes.

Until a fix arrives, the advice is to restrict access rules to specific source-to-destination pairs, consider SSH check mode, and review forwarding settings. Check mode adds a login step but does not fix the reported escalation, so it should not be treated as a patch.