CVE-2025-36320: IBM watsonx.data Intelligence Stored XSS Vulnerability

Technical summary

CVE-2025-36320 is a stored XSS vulnerability (CWE-79) in IBM watsonx.data intelligence affecting versions 5.2.0, 5.2.1, 5.2.2, and 5.3.0. The vulnerability exists in the Web UI where insufficient input validation or output encoding allows authenticated users to embed arbitrary JavaScript code. Because the payload is stored server-side, it executes in the context of any user's session that accesses the affected component, bypassing traditional client-side XSS mitigations. The CVSS 3.1 score of 6.4 (MEDIUM) reflects the network-accessible attack surface combined with the requirement for prior authentication and the confidentiality and integrity impact.

Business impact

Organizations using affected IBM watsonx.data intelligence versions face credential compromise risk within trusted user sessions. An insider or low-privilege user could inject JavaScript to harvest authentication tokens, session cookies, or sensitive data displayed in the UI. While the vulnerability requires authentication to introduce the payload, the stored nature means the attack affects all subsequent users, potentially creating widespread exposure across the enterprise. Data analytics teams relying on watsonx.data for business intelligence workflows could experience unauthorized access to dashboards, queries, and underlying data.

Affected systems

IBM watsonx.data intelligence versions 5.2.0, 5.2.1, 5.2.2, and 5.3.0 are vulnerable. IBM Software Hub products that depend on or integrate with watsonx.data intelligence may also be affected depending on deployment architecture. Organizations should verify exact product versions in their environments, as patch status varies.

Exploitability

Exploitation requires valid user credentials and interactive access to the Web UI, making opportunistic mass exploitation unlikely. However, the stored nature significantly lowers the barrier compared to reflected XSS—once injected, the payload executes automatically for all users viewing the compromised page without requiring the victim to click a malicious link. Insider threats or compromised low-privilege accounts present the highest risk. No public exploit code is known to be in circulation, and the vulnerability is not tracked in CISA's Known Exploited Vulnerabilities catalog.

Remediation

Organizations must prioritize patching to a fixed version of IBM watsonx.data intelligence. Verify availability of patches for versions 5.2.x and 5.3.0 through IBM's security advisories. Until patches are deployed, implement input validation and output encoding reviews, restrict Web UI access to trusted networks or VPN, and audit user activity logs for unusual JavaScript injection patterns in stored data or configuration fields.

Patch guidance

Consult IBM's official security advisory for CVE-2025-36320 to obtain patched versions and verify exact version numbers. Plan patching in coordination with your release management cycle, as watsonx.data intelligence may be mission-critical for analytics workflows. Test patches in a staging environment first to ensure compatibility with existing queries, integrations, and ETL processes. Document the patch status and deployment date for compliance audits.

Detection guidance

Monitor Web UI traffic and logs for stored XSS indicators: HTML entity encoding inconsistencies in configuration or data fields, presence of script tags in unexpected locations, and JavaScript execution from stored content. Review audit logs for users injecting suspicious payloads into shared dashboards, queries, or metadata. Inspect browser developer console logs and network traffic for unexpected JavaScript sources. Consider deploying a Web Application Firewall (WAF) configured to block or flag JavaScript injection attempts in POST/PUT requests to watsonx.data endpoints.

Why prioritize this

Although the CVSS score is MEDIUM (6.4), the stored XSS nature and access to sensitive data warrant attention above generic medium vulnerabilities. Authentication requirement limits blast radius but does not eliminate risk—insider threats and credential compromise are persistent concerns. Organizations should prioritize patching within their standard medium-severity timelines (typically 30–60 days), but expedite if watsonx.data is exposed to untrusted internal networks or if user base includes contractors with limited oversight.

Risk score, explained

The CVSS 3.1 score of 6.4 reflects: network-accessible attack surface (AV:N), low attack complexity (AC:L), requirement for authenticated user (PR:L), no user interaction needed once payload is stored (UI:N), and scope change allowing impact across trust boundaries (S:C). Confidentiality and integrity are both impacted (C:L, I:L) but availability is not (A:N). The score would be higher if the vulnerability allowed privilege escalation or unauthenticated access, or lower if exploitation required social engineering.

Frequently asked questions