NLTK versions earlier than 3.10.0 are affected by a path traversal flaw in the FramenetCorpusReader and NKJPCorpusReader components. The flaw allows an attacker to craft selector inputs or poisoned index state that cause these readers to parse XML files located outside the intended corpus root. By exploiting methods such as frame_by_name, doc, lu, and header with crafted parameters, an attacker can read any XML file accessible to the application, exposing sensitive information and compromising confidentiality.

The vulnerability impacts installations of NLTK that are below version 3.10.0, including the 3.9.x series. Any Python environment that imports NLTK and uses the FrameNetCorpusReader or NKJPCorpusReader to parse corpora is potentially exposed. Users of DSLs or applications that rely on NLTK for natural language processing should verify the library version and update accordingly.

The CVSS score of 8.2 indicates a high severity. The EPSS score is not available, so the exact likelihood of exploitation cannot be quantified from the data. The vulnerability is not listed in the CISA KEV catalog. Attackers would need to induce the application to use malicious selector values or manipulate the index state; the likely attack vector is local, via the application runtime or a remote exploitation path if the application accepts user supplied corpus inputs. The flaw provides an information‑disclosure risk by allowing arbitrary file reads within the application's accessible file system.