Tools AI-Found Vulnerability Register aifv-0124
Register entry
Linux networking lifetime error enables a research exploit
- CVE
- CVE-2026-98432
- Project
- Linux kernel
- Component
- inet options / multipath TCP
- Finder
- Unreleased internal Claude research model
- Autonomy
- model-assisted research with human prompting and review
- Evidence tier
- Claim outside the record
- Assigner
- Linux
- Severity
- 9.8 Critical (CVSS 3.1)
- Confidence
- CONFIRMED
What happened
Anthropic’s 27 October research account attributes discovery of a Linux networking weakness to its unreleased model, with human prompting and review.
Credit as recorded: null. The vendor describes a remote-code-execution demonstration against a server configured with a multipath TCP listener. The CVE was published on 15 October; public AI attribution followed on 27 October. The relevant kernel branches have their own fixes. Anthropic says the issue was reported to the vendor and patched; its unreleased model has no public version name. Date basis: Original CVE publication; explicit AI attribution on 27 October 2026.
Sources
- Anthropic original research account, 27 Octoberwww.anthropic.com/research/discovering-vulnerabilities-with-…
- Canonical CVE recordraw.githubusercontent.com/CVEProject/cvelistV5/main/cves/202…
One record from the AI-Found Vulnerability Register, maintained by the Security Desk.Data: CSV · JSON ·RSS · CC BY 4.0 with attribution to YFarmX.Tracker updated · 28 October 2026