Critical Linux flaw exposes self-hosted Amnezia VPN servers—patch now
Self-hosted Amnezia VPN servers are under active threat after researchers disclosed a critical Linux kernel flaw, CVE-2026-98432. The bug, rated 9.8 out of 10 for severity, lets attackers remotely seize control of vulnerable machines in under an hour—no credentials required. While Amnezia’s managed services remain safe, operators running their own servers must patch immediately or risk takeover. Xray protocol users face the highest risk.
Editor, Lazyfounder
30 SEC SUMMARY
- A critical Linux kernel flaw (CVE-2026-98432, severity 9.8/10) lets attackers remotely seize self-hosted Amnezia VPN servers in under an hour—no passwords required.
- The bug, a race condition in IPv4 TCP handling, has lurked in Linux since version 4.4 and was flagged by an Anthropic researcher after an unreleased Claude model wrote a working exploit.
- Amnezia’s Premium, Free, and Business users are safe, but self-hosted operators face a manual patching process that may not stick on some Ubuntu versions.
- Xray protocol servers are especially vulnerable due to MPTCP being enabled by default; admins must patch now, block IP options via firewall, or switch to AmneziaWG.
- The incident marks one of the first documented cases of AI generating a functional exploit, signaling a shift in how vulnerabilities are discovered and weaponized.
TABLE OF CONTENTS
A 9.8-rated Linux flaw lets attackers remotely seize Amnezia VPN servers
Amnezia VPN issued an urgent warning this week after security researchers uncovered a critical flaw in the Linux kernel, tracked as CVE-2026-98432. The vulnerability, scored 9.8 out of 10 for severity, allows attackers to remotely seize control of self-hosted servers—no passwords or user interaction required. According to TechRadar, the bug stems from a race condition in how Linux handles IPv4 TCP connections, a flaw present since kernel version 4.4.
Anthropic researcher’s finding reveals AI-written exploit
The discovery came from Xinyang Ge, a researcher at Anthropic. Ge’s report included a twist: an unreleased Claude AI model had already written a functional exploit for the flaw, capable of compromising a server in roughly 53 minutes. The exploit targets servers running the Xray protocol, which Amnezia flags as high-risk because it enables Multipath TCP (MPTCP) by default. Admins who installed Xray may have unknowingly exposed their servers to attack.
Self-hosted users face manual patching process
Amnezia’s managed services—Premium, Free, and Business—remain unaffected. The company patched its own infrastructure before issuing the public advisory. Self-hosted users, however, must act on their own. The fix involves three steps: update the Amnezia VPN app to version 5.1.0.6 or later, remove the Xray protocol, update server packages, and reinstall Xray after a reboot. The server will go offline during the process, and existing Xray connections will drop.
Patch may not persist—workarounds issued for Ubuntu users
For some Ubuntu users, the patch may not stick. On certain versions, the fix remains in the "-proposed" testing channel, meaning it could vanish after a reboot. Amnezia acknowledges the issue and offers two workarounds: block IP options via firewall rules or ditch Xray altogether and switch to AmneziaWG, a UDP-based protocol. The app now includes an update button to simplify the process, introduced after Amnezia refreshed Xray-core in September.
What this means
Lazyfounder analysis — our interpretation, not reported fact.
AI-generated exploits are no longer theoretical. Founders building security tools now face attackers who can automate the discovery and exploitation of zero-days, compressing response windows from weeks to hours. For self-hosted VPN operators, this flaw turns routine server maintenance into an urgent security drill.
Key takeaways
- Self-hosted Amnezia VPN servers using the Xray protocol remain exposed until admins complete a manual, multi-step patch—leaving a window of at least 53 minutes for attackers to strike.
- Some Ubuntu users may need to reapply the patch after every reboot until Canonical pushes a permanent fix, making temporary firewall rules or protocol switches critical stopgaps.
- Anthropic researcher Xinyang Ge’s discovery—and the unreleased Claude model’s exploit—show a new era where AI doesn’t just find flaws but weaponizes them at scale.
Related on Lazyfounder
Sources
- TechRadar · 2026-11-11
Running your own Amnezia VPN server? You need to patch this critical Linux flaw now
This story is an original summary drafted with AI by Lazyfounder from the reporting listed above and checked by automated validation. Facts are attributed to their original publishers; sections marked as analysis are Lazyfounder's. Where a source is in another language, facts were machine-translated and quotations are reported, not reproduced. Read the original coverage via the links, and see our AI policy and corrections policy.
About the author
Editor, Lazyfounder
Tarun Mottlia edits LazyFounders, covering Indian startups, funding rounds, AI and product launches. Every story on the site is AI-assisted and checked against its cited sources before publication.
More stories by Tarun MottliaGet the LazyFounder Brief
Startup, funding and AI news in a five-minute read. Join the early-access list.
