NLTK Path Traversal via Model-Artifact APIs

Executive Summary

CVE-2026-81726 is a path traversal vulnerability in NLTK versions up to 3.10.3. It allows attackers to bypass path security restrictions by using raw file operations on caller-controlled paths in certain model-artifact APIs. This enables reading or writing files outside the intended sandbox directories even when path security is enabled.

Useful 0 Not Useful 0

Detection Guidance

Check for NLTK versions <=3.10.3 using pip show nltk or pip list. Review code using vulnerable APIs like TransitionParser.train, AveragedPerceptron.save, or PerceptronTagger.save_to_json for improper path handling. Monitor file system access logs for unexpected reads/writes outside intended directories.

Useful 0 Not Useful 0

Impact Analysis

This vulnerability could allow attackers to access or modify sensitive files on your system if you use vulnerable NLTK APIs with pathsec enabled. Attackers might read configuration files, overwrite critical system files, or steal data by exploiting raw file operations in TransitionParser, AveragedPerceptron, PerceptronTagger, or maxent parameter APIs.

Useful 0 Not Useful 0

Compliance Impact

This vulnerability could lead to unauthorized data access or modification, violating compliance requirements for GDPR (data protection) and HIPAA (health information security). Unauthorized file access may result in data breaches, exposing sensitive personal or health information, which could lead to regulatory penalties and loss of trust.

Useful 0 Not Useful 0

Mitigation Strategies

Upgrade NLTK if patches are available. Replace vulnerable API calls with nltk.pathsec.open() or pathsec-aware helpers. Restrict model paths to sandbox directories. Add regression tests to validate path security. Disable pathsec enforcement only as a last resort.

Useful 0 Not Useful 0