Claude research model wrote a root exploit for Linux CVE-2026-98432
Original title: Claude AI Exploits Linux Kernel Flaw to Execute Root Commands Without User Interaction
Anthropic calls it a lab test, not a real incident; upstream kernels are patched but several Ubuntu releases remain vulnerable.
Anthropic’s Frontier Red Team said on October 27 that an unreleased Claude model produced an exploit for CVE-2026-98432 after one researcher prompt. The company describes a race and use-after-free on servers with a multipath TCP listener, leaking a kernel pointer and then running a command as root. Ubuntu lists CVSS 9.8 and describes an IPv4 options bug fixed with kfree_rcu(). Anthropic says the fix is upstream and backported to maintained stable branches.
Unconfirmed: No outside team has confirmed Anthropic’s chain, and the Ubuntu record never mentions multipath TCP.
- Ubuntu published the CVE entry on October 15 and updated it October 28.
- A blog lists fixes in 5.10.272, 5.15.223, 6.1.190, 6.6.159, 6.12.113 and 6.18.56.
- Ubuntu 22.04, 24.04 and 26.04 were still marked vulnerable, with work in progress.
- The proof of concept took 48 minutes to leak a pointer, then five more.
- The Ubuntu write-up does not mention multipath TCP.
Impact: remote command execution as root without login
| Product | Affected | Fixed in |
|---|---|---|
| Linux kernel | — | 5.10.272, 5.15.223, 6.1.190, 6.6.159, 6.12.113, 6.18.56 |
| Canonical Ubuntu kernel | 22.04, 24.04, 26.04, 20.04 and 18.04 marked vulnerable | — |
- 1.Install the latest distribution kernel update and reboot.
- 2.Disable unneeded multipath TCP listeners until a patch is installed.
- 3.Confirm the listed stable fix versions on kernel.org.
- CVEs
- CVE-2026-98432
- Vendors
- Canonical
- Products
- Linux kernelUbuntu
- Organizations
- Anthropic
- AI models
- Claude
Vulnerabilities mentioned
All| CVE | Vulnerability | CVSS | EPSS | Signals | Published |
|---|---|---|---|---|---|
| CVE-2026-98432 | Use-after-free in Linux kernel IPv4 TCP options | 9.8 | <1% | PoC ×15 |
From the original article
4 min read at cybersecuritynews.com
Anthropic says an unreleased Claude research model wrote a working exploit for a critical Linux kernel bug, tracked as CVE-2026-98432. The exploit lets a remote attacker run commands as root with no login and no user action. Anthropic’s Frontier Red Team shared the finding on October 27, 2026, and the flaw is already patched upstream.
By Guru Baran · cybersecuritynews.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.