NLTK CVE-2026-79674

Exploitation requires that the calling application pass externally controlled input as the corpus root path argument to LinThesaurusCorpusReader or PanLexLiteCorpusReader; NLTK corpus readers that do not accept external root paths, or applications that hardcode the corpus root to a fixed value, are not exploitable via this vulnerability. The CVSS 4.0 vector (PR:N/UI:N/AV:N) applies only to deployments where these constructors are reachable from an untrusted network context - this is a non-default application design pattern requiring explicit exposure. Exploitation is bounded by the OS-level permissions of the NLTK process: only files readable by that process account can be accessed, and privilege escalation beyond the process's existing OS permissions is not described. The pathsec bypass targets specifically the two named constructors; other NLTK components are not described as affected.

The CVSS 4.0 score of 8.8 (AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:L/VA:N) reflects the worst-case scenario: a network-accessible application that exposes the affected constructors to unauthenticated user-supplied input. High confidentiality impact (VC:H) is well-supported by the ability to read arbitrary files and SQLite databases accessible to the NLTK process; the low integrity impact (VI:L) is not fully explained by the description, which emphasizes read access, and should be independently verified against the upstream advisory. EPSS data was not provided in the available intelligence, so exploitation probability cannot be quantified numerically. No active exploitation is confirmed - neither CISA KEV nor a public POC is present in the sourced data. Practically, NLTK is predominantly deployed in offline or internal data-science environments where corpus-reader constructors are not exposed to external input, which would significantly reduce real-world exploitability below what the raw CVSS score implies; however, any deployment embedding NLTK in a web API that accepts corpus path parameters without sanitization faces genuine high-severity risk.

An attacker identifies a web API that accepts a corpus name or root path parameter and forwards it directly to PanLexLiteCorpusReader or LinThesaurusCorpusReader. By supplying a crafted path such as an absolute path to /etc/shadow or a relative traversal sequence targeting application configuration files or credential databases, the attacker causes NLTK to open and return content from outside the intended data directory. No public POC has been identified at time of analysis, but the attack requires no specialized tooling beyond a standard HTTP client capable of sending a crafted request.