CVE-2026-126304

A corpus could lead the NLTK reader into another dataset

NLTK project · Natural Language Toolkit York-Toronto-Helsinki corpus reader

First documented disclosure · Review checkpoint

Summary

The reader followed links in corpus subdirectories when accessing parsed or tagged text. The reader could expose files outside the intended corpus but still within the broader NLTK data sandbox.

Severity and exploitation evidence

High · 8.7/10 · CVSS 4.0. Severity source.

Exploitation not established: The reviewed advisory does not report malicious exploitation.

Exploitation evidence source

Technical severity is not a measure of observed attacks or of risk to your deployment. Attempts, demonstrations, and successful compromise retain the qualifications in the source.

How the weakness works

The reader followed links in corpus subdirectories when accessing parsed or tagged text.

Conditions required

An attacker controlled a YCOE corpus read by the application.

Potential impact

The reader could expose files outside the intended corpus but still within the broader NLTK data sandbox.

Affected versions and configurations

3.10.0 through 3.10.3.

Fix and mitigation status

3.10.4.

Practical next steps

Upgrade NLTK to 3.10.4 or later. Isolate externally supplied corpora from other tenants’ data and shared temporary files.

Operating lesson

Apply a separate filesystem boundary to each corpus.

Evidence and limits

The CVE record was published on 2026-11-13.

Sources