# Anthropic’s Claude writes Linux exploit for unpatched CVE-2026-98432

Digest AI · Society & Work · published 2026-11-02T04:04:00Z

Canonical: https://digestai.news/story/anthropics-claude-writes-linux-exploit-for-unpatched-cve-2026-98432

## Summary

Anthropic disclosed on October 27 that its AI model generated a working exploit for **CVE-2026-98432**, a critical Linux vulnerability scoring **9.8/10** in severity. The bug, present since Linux **4.4**, lets attackers gain full server control without passwords by exploiting a race condition in memory handling. Anthropic’s exploit uses an unreleased internal Claude model and targets systems with **multipath TCP** enabled, though the underlying flaw affects other configurations too.

Ubuntu’s tracker still marks versions **16.04** through **26.10** as vulnerable, with fixes only confirmed for **22.04**, **24.04**, **26.04**, and **26.10**. Red Hat, Debian, SUSE, Oracle Linux, and Amazon Linux have released patches for some versions, but Ubuntu has not yet issued fixes for **20.04**, **18.04**, or **16.04**. A temporary workaround from Red Hat reduces exposure but does not fully block attacks. Anthropic previously used its **Claude Mythos** model to help Mozilla discover **271 bugs** in Firefox.

## Key points

- Anthropic’s AI wrote an exploit for **CVE-2026-98432**, a Linux vulnerability scoring **9.8/10** in severity
- Attackers can gain root access in under **1 hour** by exploiting a memory race condition in **multipath TCP**
- Ubuntu still lists **16.04–26.10** as vulnerable, with fixes only confirmed for **22.04**, **24.04**, **26.04**, and **26.10**

## Why it matters

A working exploit for an unpatched Linux flaw demonstrates how AI-generated attacks could bypass traditional defenses, raising concerns for servers still running vulnerable versions.

## Sources

1. [Claude-Built Linux Exploit Targets Bug That Ubuntu Still Hasn't Patched](https://tech.yahoo.com/ai/claude/articles/claude-built-linux-exploit-targets-203603880.html) (tech.yahoo.com, 2026-11-02)

Part of the developing story: [Claude AI Discovers and Exploits Linux Kernel Flaw](https://digestai.news/thread/anthropic-s-unreleased-claude-ai-finds-critical-linux-kernel-flaw-cve-2026-98432) (2 stories)

## Cite

Digest AI, "Anthropic’s Claude writes Linux exploit for unpatched CVE-2026-98432", 2 November 2026, https://digestai.news/story/anthropics-claude-writes-linux-exploit-for-unpatched-cve-2026-98432

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/anthropics-claude-writes-linux-exploit-for-unpatched-cve-2026-98432.json