Impact
NLTK versions prior to 3.10.3 allow a path sandbox bypass in corpus‑reader constructors, which can result in reading files outside the intended data root.
Affected Systems
Any deployment using NLTK before 3.10.3 is affected, with specific impact on the LinThesaurusCorpusReader and PanLexLiteCorpusReader components. Versions 3.10.3 and later include the fix.
Risk and Exploitability
The CVSS score of 8.8 categorizes this as high severity. The EPSS score of < 1% indicates that exploitation probability is very low, and the issue is not listed in the CISA KEV catalog. Attackers can exploit the flaw by supplying a crafted corpus root path; the requirement for an attacker to run Python code or influence the corpus path suggests the threat is most relevant to local or untrusted environments where unvalidated input reaches the reader constructors.
OpenCVE Enrichment
Github GHSA