The Tailscale SSH fix is 1.104.4
Tailscale 1.104.4 fixes four SSH and Serve bulletins. Exploitation needs an authorized foothold. The company says it is not a self-replicating worm.
The short version
No. Tailscale says an attacker needs an existing foothold on a policy-authorized device. Version 1.104.4 fixes TS-2026-014 through TS-2026-017.
The company post
Tailscale published Tailscale SSH: Findings and fixes with version 1.104.4. Four security bulletins cover the confirmed issues: TS-2026-014 through TS-2026-017. The company says the problems were about privilege boundaries after a connection was already authorized, not about skipping authentication. Exploitation requires a foothold on a policy-authorized peer. The Hacker News summarized the same version and the same foothold requirement. This page will not describe forwarding paths as a how-to. The document named in the sources is the record this page will keep. The document named in the sources is the record this page will keep.
Not a worm, on Tailscale's wording
Tailscale says the findings are better described as lateral movement limited by the tailnet policy file, not as a computer worm. That sentence is the company's. A reader who only sees the research nickname can miss it. PQ GeoKey was a closed beta of a different networking idea. Origin Range was a cache rule. Neither one is Tailscale SSH. The document named in the sources is the record this page will keep. The document named in the sources is the record this page will keep.PQ GeoKey.
Who is in scope
Operators who enable Tailscale SSH, especially with local or remote forwarding allowed, and nodes that use Serve features covered in the bulletins. People who never enable Tailscale SSH are outside the SSH bulletins. Reviewing SSH policy is the company's additional advice beside the upgrade. Asked directly, What interim control does Tailscale name? Setting TS_SSH_DISABLE_FORWARDING=1 in the tailscaled environment and restarting the service, which disables forwarding without turning off Tailscale SSH. The document named in the sources is the record this page will keep. The document named in the sources is the record this page will keep.A name is not a location.
The interim flag
Tailscale names TS_SSH_DISABLE_FORWARDING=1 as a mitigation that turns off forwarding without disabling Tailscale SSH. Naming the flag is the vendor's documented interim control. This page will not expand it into a distribution-specific unit file tour. The advisory habit still applies: the vendor text wins. Asked directly, Who should update first? Nodes that run Tailscale SSH or Tailscale Serve, especially where forwarding is allowed in policy. The document named in the sources is the record this page will keep. The document named in the sources is the record this page will keep.The advisory is the story.
headscale is called out in news accounts
The Hacker News notes that headscale's newest release did not fix the reported issues and that forwarding behavior there differs. That is an account of the research report. Tailscale's hosted bulletins are the fix path this brief tracks. A separate headscale filing would need a headscale release note. The document named in the sources is the record this page will keep. The document named in the sources is the record this page will keep. The document named in the sources is the record this page will keep.Origin Range.
Microsoft's Shell bypass is Windows
Today's Microsoft brief is a Windows Mark of the Web bypass. Tailscale's brief is a mesh SSH privilege boundary. Both are October security-adjacent documents. Only one belongs on the networking desk. Asked directly, What does Tailscale say is required to exploit? Access to an authorized device in the tailnet, with SSH access permitted in the policy, plus the conditions needed for the vulnerable paths. The document named in the sources is the record this page will keep. The document named in the sources is the record this page will keep.October's Shell bypass.
What this page will not paste
Socket names, lab chain timings, and any sequence that turns a foothold into root stay in the research writeups and out of this brief. The version, the bulletin IDs, the foothold requirement, and the non-worm sentence are the dependency. Asked directly, Does Tailscale call the findings a worm? No. The company says the report name prompted that question and that the issues are not self-replicating malware between machines. The document named in the sources is the record this page will keep. The document named in the sources is the record this page will keep.
Node October stays a runtime
The Node October patch was a language runtime. Tailscale 1.104.4 is a mesh client. Linking them only separates the desks. The short answer this page will stand on is this: No. Tailscale says an attacker needs an existing foothold on a policy-authorized device. Version 1.104.4 fixes TS-2026-014 through TS-2026-017. The document named in the sources is the record this page will keep. The document named in the sources is the record this page will keep. The document named in the sources is the record this page will keep.The Node October patch.
The record
The record is Tailscale's findings post, the four bulletin pages, and The Hacker News account. Fixed in 1.104.4. Foothold required. Not a worm, the company says. Asked directly, Which Tailscale version fixes the SSH findings? 1.104.4 or later. Tailscale published four bulletins, TS-2026-014 through TS-2026-017. The document named in the sources is the record this page will keep. The document named in the sources is the record this page will keep. The document named in the sources is the record this page will keep. The document named in the sources is the record this page will keep.
Sources
The reports this brief is filing. Futureweb did not republish them.
Questions
Which Tailscale version fixes the SSH findings?
1.104.4 or later. Tailscale published four bulletins, TS-2026-014 through TS-2026-017.
What does Tailscale say is required to exploit?
Access to an authorized device in the tailnet, with SSH access permitted in the policy, plus the conditions needed for the vulnerable paths.
Does Tailscale call the findings a worm?
No. The company says the report name prompted that question and that the issues are not self-replicating malware between machines.
What interim control does Tailscale name?
Setting TS_SSH_DISABLE_FORWARDING=1 in the tailscaled environment and restarting the service, which disables forwarding without turning off Tailscale SSH.
Who should update first?
Nodes that run Tailscale SSH or Tailscale Serve, especially where forwarding is allowed in policy.