IBM watsonx.ai cross-site scripting vulnerability CVE-2024-49785
IBM disclosed a cross-site scripting vulnerability in watsonx.ai that allows authenticated users to inject malicious JavaScript code through third-party LLM prompts in the Web UI, potentially leading to credential disclosure. IBM released patched versions to address the flaw.
What happened
IBM watsonx.ai for Cloud Pak for Data contains a cross-site scripting (XSS) vulnerability (CVE-2024-49785) in its Web UI interface when using unauthorized third-party large language model (LLM) prompts.
An authenticated user can embed arbitrary JavaScript code in the Web UI, potentially altering intended functionality and leading to credentials disclosure within a trusted session. The vulnerability has a CVSS Base score of 5.4.
Reported impact
An authenticated user could inject malicious JavaScript code to disclose credentials or alter the application's intended functionality during a trusted session.
Index analysis
The vulnerability affects IBM watsonx.ai versions 1.1 through 2.0.3, and IBM watsonx.ai on Cloud Pak for Data versions 4.8 through 5.0.3.
Governance controls
IBM released fixed versions: watsonx.ai 2.1.0 and above, and watsonx.ai on IBM Software Hub 5.1.0 and above. No workarounds were available and IBM strongly recommended upgrading to patched versions.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.