CVE-2021-44550: Access Control vulnerability within CoreNLP
9.8
CVSS Score
3.1
Basic Information
CVE ID
GHSA ID
EPSS Score
0.62825%
CWE
Published
2/25/2022
Updated
2/3/2023
KEV Status
No
Technology
Java
Technical Details
CVSS Vector
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
Package Name | Ecosystem | Vulnerable Versions | First Patched Version |
---|---|---|---|
edu.stanford.nlp:stanford-corenlp | maven | <= 4.3.2 | 4.4.0 |
Vulnerability Intelligence
Miggo AI
Root Cause Analysis
The vulnerability stems from NERServlet.java
's addResults()
method adding response headers (lines 158-159) before validating the 'classifier' parameter. The original code: 1) took user input from request.getParameter('classifier')
, 2) immediately set it in response headers, and 3) only later used it for model lookup. This allowed attackers to inject arbitrary headers and potentially access unauthorized models by manipulating the classifier parameter. The patch fixed this by first validating the classifier exists in the model map before setting headers.