IBM Engineering AI Hub CVE-2026-15091

Exploitation requires that a victim with access to IBM Engineering AI Hub (versions 1.0.0-1.2.0) interact with attacker-controlled content - this is the UI:R limiting factor: the attacker cannot fire the payload autonomously and must get a user to load a crafted request, link, or page where unsanitized input is rendered. No authentication or privileges are needed on the attacker side (PR:N), and the vector is network-reachable (AV:N) against the web interface, but the requirement for user interaction and access to the web-facing application constrains exploitation to scenarios where an attacker can reach and socially engineer a valid user. The impact is bounded to what script can do within the victim's authenticated browser context.

The vendor CVSS 3.1 vector (AV:N/AC:L/PR:N/UI:R/S:C/C:H/I:H/A:N, 9.3) rates this Critical, driven largely by C:H/I:H combined with the changed scope - an aggressive impact rating for XSS, which typically confers Low confidentiality/integrity impact bounded by the victim's browser context; verify the High impact rating against IBM's advisory before treating it as full-session compromise. Attack complexity is Low and no privileges are required (PR:N, unauthenticated), but exploitation is gated by required user interaction (UI:R), meaning an attacker must lure or wait for a victim to trigger the payload. No EPSS score, no POC, and no KEV listing were provided, so there is no evidence of active exploitation or automatable mass-exploitation; real-world urgency is moderated by the UI:R dependency and the niche deployment footprint despite the high nominal score.

An attacker crafts a malicious URL or input that embeds a JavaScript payload and delivers it to a legitimate Engineering AI Hub user (e.g. via a phishing link or a shared engineering artifact). When the victim opens it, the application reflects/renders the unsanitized input and the browser executes the script in the victim's authenticated session, allowing session token theft, UI manipulation, or actions performed as the victim. No public exploit code was identified; the AV:N/AC:L/UI:R profile means exploitation is technically easy but depends on victim interaction.