CVCOVANewsComputing, Opportunities, Value & Analysis
Cybersecurity & Privacy5 min read

A critical flaw in Linux networking: which servers are exposed and how to patch them

The vulnerability CVE-2026-98432, scored 9.8 out of 10, reportedly affects the Linux kernel going back to 2016. Debian, SUSE, Oracle and Amazon have already released fixes; Red Hat and Ubuntu still had none as of late October.

By COVA News newsroom
A critical flaw in Linux networking: which servers are exposed and how to patch them

Every time a server receives a visit, it all begins with a greeting: a small packet called SYN, the first message one computer sends to ask to open a connection with another. Hidden in that routine gesture, which happens millions of times per second across the internet, is CVE-2026-98432, a vulnerability in the Linux kernel (the core of the operating system, the component that controls memory, processes and networking). It received a severity score of 9.8 out of 10 and has spent days among the most upvoted security topics on Hacker News, with 643 points and 311 comments.

The team responsible for cataloging kernel vulnerabilities published the record on October 15, 2026, and added the CVSS score (the standard scale for measuring a flaw's severity) on October 24. According to the published technical description, the problem arises when a server receives a SYN carrying "IP options" (optional fields in packet headers) sent to an open TCP port that is listening for connections. The attack would be carried out remotely, over the network, without passwords and without anyone having to click on anything. That explains such a high score.

The flaw belongs to a family well known to specialists: a *use-after-free* triggered by a race condition. Put simply, two processes compete for the same chunk of memory and, if events happen in the wrong order, the system keeps using a space it had already freed. A skilled attacker can exploit that disorder to place their own data there. According to AlmaLinux, the cybersecurity firm Rescana and researcher Pasquale Pillitteri, the defect has reportedly been present since Linux 4.4, a 2016 release. It would affect servers, appliances (network devices with Linux inside) and also containers, because they share the kernel of the host machine.

An AI-built exploit that no one has been able to review

Interest grew because of an unusual ingredient. Anthropic says an as-yet-unreleased Claude model generated an exploit, that is, a program that takes advantage of the flaw, capable of remotely gaining full ("root") control of servers that use MPTCP, a variant of TCP that lets a single connection travel over several network paths at once. A report by researchers dated October 27 describes that lab attack chain, which reportedly achieved control in about 53 minutes.

This should be read with caution. The code, kernel and configuration used are not public, and no one has independently reproduced the result. The official CVE record doesn't even mention MPTCP, so it is not confirmed that a server without that feature allows remote code execution. Even so, the case shows how artificial intelligence is beginning to move into vulnerability research.

Nor is there consensus on whether the flaw is already being used in real-world attacks. The vendor Aviatrix maintains that there are active attacks and a publicly disclosed exploit, but provides no evidence. TuxCare, Rescana and Cybolt's commercial CSIRT say the code is not public and no exploitation is known. The US agency CISA's catalog of exploited vulnerabilities did not include it in its October 29 version, and the firm Ostorlab confirmed the same on November 1.

Severity also varies depending on whom you ask. Red Hat raised it to Critical (9.8). According to a blog, Oracle and SUSE rate it 8.1 and Amazon between 7.0 and 8.1, because they consider the attack more difficult to carry out.

Who already has a patch and who is still waiting

The fix is in the mainline kernel code (commit e3f33b0a1d89). According to Rescana, the fixed stable versions are 5.10.272, 5.15.223, 6.1.190, 6.6.159, 6.12.113, 6.18.56, 7.2.10 and 7.3. Here is where each distribution stood according to the advisories reviewed:

  • Debian: fixed. Debian 13 with linux 6.12.113-1 (DSA-6568-1) and Debian 12 with 6.1.190-1 (DLA-4841-1) or linux-6.12 6.12.113-1~deb12u1 (DLA-4850-1).
  • SUSE: fixed in SLES 16.0, SLE Micro 6.2 and openSUSE Leap 16.0 (kernel-default 6.12.0-160000.40.1 or later), as well as SLES 15 SP4–SP6 LTSS and SLE Micro 5.3–5.5. SLES 15 SP7 was still pending.
  • Oracle Linux UEK: fixed since October 16 (ELSA-2026-500397/8/9).
  • Amazon Linux 2023: fixed (ALAS2023-2026-3251 and -3335).
  • AlmaLinux 8, 9 and 10: affected; the patched kernels were available only in the testing repository as of October 30.
  • RHEL 8, 9 and 10 and OpenShift: affected, with no fixed kernel. RHEL 6 and 7 are not affected.
  • Ubuntu (16.04 through 26.10): no fixed kernel and no release date as of October 31.

For those without a patch, Red Hat published bulletin RHSB-2026-016 on October 30 with a packet-filtering rule for nftables, the Linux firewall. The company itself warns that it reduces exposure but does not fully prevent exploitation.

In Latin America, where RHEL, Ubuntu, Debian and AWS are common in banking, telecommunications and startups, the immediate task is to take inventory. On systems with a fix available, update the kernel and reboot, because the patch doesn't take effect until the machine boots into the new version. On RHEL and Ubuntu, administrators should evaluate Red Hat's mitigation and keep an eye on official advisories. So far, no alerts have been found from CERT.br, Chile's CSIRT, ColCERT or CERT.ar. Spanish-language advisories were issued by the Universidad Católica Boliviana's Tarija campus and by Cybolt. It remains to be seen when the Red Hat and Ubuntu kernels will arrive, who discovered the flaw, and whether servers without MPTCP face the same risk.

Related news

Cybersecurity & Privacy

An official Microsoft package stole credentials on PyPI: what to check if you use AI agents

Two versions of Microsoft's agent-framework reached PyPI carrying a credential stealer, one day after Google warned of attackers using AI to exploit flaws without human intervention. Here is what development and IT teams in Latin America should check right now.

Cybersecurity & Privacy

Microsoft and FAA outages, FBI and USDA breaches: no evidence of a coordinated attack

Over nine days, service outages and alleged intrusions coincided in the United States, and social media buzzed with talk of a “cyber pandemic.” The evidence points to separate technical failures. The proven danger for users is something else: fake emails that use the outages as bait.

Cybersecurity & Privacy

Nginx: two critical HTTP/2 flaws expose millions of servers; here's how to check and protect yours

F5 has fixed two vulnerabilities in Nginx that allow an attacker to take control of a server without a username or password. Public proof-of-concept code is already circulating. This guide explains how to tell whether your server is at risk, which version to update to and what to do while you wait for your system's patch.