CVE-2025-36321 Overview
CVE-2025-36321 is an HTML injection vulnerability affecting IBM watsonx.data intelligence versions 5.2.0, 5.2.1, 5.2.2, and 5.3.0. A remote authenticated attacker can inject malicious HTML markup into the application. When a victim views the injected content, the browser renders it within the security context of the hosting site. This vulnerability is classified under CWE-80: Improper Neutralization of Script-Related HTML Tags in a Web Page. IBM assigned this issue an EPSS score of 0.398% with a percentile of 31.787, indicating low observed exploit activity to date.
Critical Impact
Successful exploitation allows attackers to inject arbitrary HTML content that executes in a victim's browser under the trust boundary of the IBM watsonx.data intelligence application, enabling phishing, credential theft, and session manipulation.
Affected Products
- IBM watsonx.data intelligence 5.2.0
- IBM watsonx.data intelligence 5.2.1
- IBM watsonx.data intelligence 5.2.2
- IBM watsonx.data intelligence 5.3.0
Discovery Timeline
- 2026-06-30 - CVE-2025-36321 published to NVD
- 2026-07-01 - Last updated in NVD database
Technical Details for CVE-2025-36321
Vulnerability Analysis
CVE-2025-36321 stems from improper neutralization of HTML markup in user-controlled input processed by IBM watsonx.data intelligence. The application accepts input containing HTML tags and renders that content back to other users without sufficient encoding or sanitization. When a victim loads the affected view, the browser interprets the injected markup as legitimate site content. Attackers leverage this behavior to embed deceptive forms, hyperlinks, or iframes into trusted pages. The requirement for user interaction and low-level privileges limits mass exploitation but does not prevent targeted attacks against authenticated users.
Root Cause
The root cause is missing or incomplete output encoding of user-supplied fields rendered in the web interface. The application fails to enforce a strict allowlist for HTML entities before injecting content into the response body. This aligns with CWE-80, which addresses improper handling of script-related HTML tags. Refer to the IBM Support Page for vendor-provided technical details.
Attack Vector
An authenticated remote attacker submits HTML payloads through an input field exposed by watsonx.data intelligence. The payload is stored or reflected and later rendered when another user views the affected page. Because the injected markup executes in the security context of the hosting site, attackers can build phishing overlays, redirect users to attacker-controlled hosts, or manipulate the DOM to capture sensitive data. The vulnerability does not directly compromise integrity or availability of the platform but places confidentiality of user session data at risk.
The vulnerability manifests through unsanitized rendering of HTML entities. See the IBM Support Page for vendor guidance and available fixes.
Detection Methods for CVE-2025-36321
Indicators of Compromise
- Web application log entries containing HTML tag sequences such as <iframe>, <img onerror=>, or <a href=javascript:> submitted to watsonx.data intelligence input fields.
- Unexpected outbound requests from user browsers to unfamiliar domains initiated while viewing watsonx.data intelligence pages.
- User reports of unfamiliar login prompts, redirects, or UI overlays appearing inside the watsonx.data intelligence interface.
Detection Strategies
- Deploy a web application firewall (WAF) rule set that inspects POST and PUT bodies destined for watsonx.data intelligence endpoints for HTML tag patterns.
- Correlate authentication events with content submission events to identify low-privilege accounts writing HTML-tagged payloads into stored fields.
- Review browser Content Security Policy (CSP) violation reports for blocked inline content originating from watsonx.data intelligence pages.
Monitoring Recommendations
- Enable verbose application logging on watsonx.data intelligence endpoints that accept free-form user input.
- Forward web server, reverse proxy, and application logs to a centralized SIEM for pattern-based analysis of HTML injection attempts.
- Alert on repeated failed input validation events tied to a single authenticated user session.
How to Mitigate CVE-2025-36321
Immediate Actions Required
- Apply the security update referenced in the IBM Support Page for watsonx.data intelligence versions 5.2.0, 5.2.1, 5.2.2, and 5.3.0.
- Audit stored records in watsonx.data intelligence for existing HTML markup and remove malicious content before end users are exposed.
- Reduce the number of accounts with write access to shared watsonx.data intelligence content until patches are deployed.
Patch Information
IBM has published remediation guidance for CVE-2025-36321 on the IBM Support Page. Administrators should consult the vendor advisory for fixed version numbers and upgrade instructions applicable to their deployment.
Workarounds
- Enforce a strict Content Security Policy (CSP) that disables inline scripts and restricts frame sources on watsonx.data intelligence hosts.
- Place an application-layer proxy in front of watsonx.data intelligence to strip HTML tags from request bodies where the field is expected to contain plain text.
- Restrict access to input surfaces to trusted internal networks until the patch is applied.
# Example CSP header to reduce HTML injection impact
Content-Security-Policy: default-src 'self'; script-src 'self'; frame-src 'none'; object-src 'none'; base-uri 'self'
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
