Apache OpenNLP’s DictionaryEntryPersistor initializes a SAXParserFactory that allows external entity resolution and DOCTYPE declarations, creating an XXE vulnerability (CWE-611). When the public Dictionary(InputStream) constructor processes a user‑supplied dictionary file, an attacker can craft a malicious DOCTYPE to read local files via file:// references or trigger server‑side request forgery using http:// references. This leads to disclosure of sensitive files or internal resources before any dictionary entry is processed. The impact is a compromise of confidentiality.

The vulnerability affects Apache OpenNLP versions prior to 2.5.9 and prior to 3.0.0-M3. The affected vendor is the Apache Software Foundation for its OpenNLP library. Users running these older releases and loading dictionaries through the public API are at risk.

The EPSS score is < 1% and the vulnerability is not listed in the CISA KEV catalog, but the lack of a secure parser makes the flaw exploitable when a trusted dictionary source is not enforced. Exploitation requires the attacker to supply a crafted dictionary file, which is realistic for deployments that load user‑supplied dictionaries. The CVSS score of 9.1 indicates high severity, indicating that successful exploitation can lead to significant confidentiality compromise.