IBM Engineering AI Hub Cross-Site Scripting Vulnerability (CVE-2026-15091)

Technical Analysis

CVE-2026-15091 stems from improper neutralization of user-supplied input during dynamic web page generation in IBM Engineering AI Hub, a platform used to manage and orchestrate AI model workflows and agent pipelines. Attackers can craft malicious payloads (e.g., embedded <script> tags or event handlers) delivered via URLs, form fields, or API responses rendered in the Hub's UI, resulting in stored or reflected XSS execution in victims' browsers. Because AI Hub is used to configure and monitor AI agents, RAG pipelines, and model integrations, a successful XSS attack could allow adversaries to hijack administrator sessions, exfiltrate API keys and service credentials used by connected agents, or silently modify agent configurations and permissions. The 9.3 CVSS score reflects the network-based attack vector, low complexity, and high impact on confidentiality, integrity, and availability. This vulnerability directly threatens organizations running AI agents through IBM Engineering AI Hub, as compromised sessions could be leveraged to manipulate agent behavior, exfiltrate connected API keys, or pivot into downstream automated systems.

Affected Systems

IBM Engineering AI Hub versions 1.0.0, 1.1.0, and 1.2.0 running on supported enterprise deployments (on-premises and hybrid cloud configurations)

Indicators of Compromise

• N/A - no public IOCs reported at this time; monitor for anomalous script injection payloads in AI Hub logs
• Suspicious URL parameters containing <script> or javascript: strings
• Unexpected outbound connections from AI Hub web sessions to unknown domains

Remediation Steps

1. 1

   Apply Vendor Patch

   Upgrade IBM Engineering AI Hub to the latest patched version as specified in IBM's security advisory once released.

2. 2

   Input Sanitization Review

   Implement or verify server-side input validation and output encoding for all user-controllable fields rendered in the AI Hub web interface.

3. 3

   Deploy Content Security Policy (CSP)

   Enforce strict CSP headers to prevent execution of unauthorized inline scripts in the web application.

4. 4

   Session and Credential Hardening

   Rotate API keys and session tokens used by AI agents integrated with AI Hub, and enforce short-lived, scoped credentials.

5. 5

   Web Application Firewall (WAF) Rules

   Deploy WAF rules to detect and block common XSS payload patterns targeting AI Hub endpoints until patching is complete.

6. 6

   Monitor and Audit

   Enable enhanced logging and monitoring of AI Hub administrative sessions and agent configuration changes to detect exploitation attempts.