CVE-2026-79674: NLTK 3.10.2 Path Traversal via corpus-reader constructors
NLTK versions before 3.10.3 contain a path sandbox bypass vulnerability in corpus-reader constructors that allows attackers to read files outside the intended data root. Attackers can supply arbitrary corpus root paths to LinThesaurusCorpusReader and PanLexLiteCorpusReader constructors to access filesystem content and SQLite databases outside the pathsec sandbox boundary.
Affected Software
Remediation
Recommended actions to resolve this vulnerability, in priority order.
- Upgrade
Upgrade to a fixed release to a version that resolves this vulnerability.
Fixed in 3.10.3 - Compensating control
Ensure applications using NLTK corpus-reader constructors do not accept attacker-controlled values for corpus root paths (e.g., validate/whitelist paths so they cannot traverse outside the intended pathsec sandbox boundary).
Event History
Frequently Asked Questions
Who is exposed to this issue?
Applications using NLTK versions before 3.10.3 are exposed if untrusted input can control the corpus root path passed to LinThesaurusCorpusReader or PanLexLiteCorpusReader. The issue can expose files and SQLite databases outside the intended NLTK data-root sandbox.
Does exploitation require authentication or user interaction?
No. The supplied vector indicates network-reachable exploitation with low complexity, no privileges required, and no user interaction, provided an attacker can supply an arbitrary corpus root path to an affected constructor.
What is the recommended remediation?
Upgrade NLTK to version 3.10.3 or later. If an immediate upgrade is not possible, do not allow untrusted input to determine corpus root paths for the affected constructors, and restrict those paths to an approved data root.
How can I determine whether an application is affected?
Check whether it uses an NLTK version earlier than 3.10.3 and instantiates LinThesaurusCorpusReader or PanLexLiteCorpusReader. It is particularly affected if corpus root paths originate from requests, configuration controlled by untrusted users, uploaded content, or other attacker-influenced sources.