CVE-2026-117494

Sisense AI assistant renders script-bearing SVG images

Sisense · Cloud BI Platform AI assistant

First documented disclosure · Review checkpoint

Summary

An SVG submitted to the AI sessions API can execute JavaScript when rendered in the assistant window.

Severity and exploitation evidence

High · 8.3/10 · CVSS 4.0. Severity source.

Exploitation not established: Real-world malicious exploitation is not established by the reviewed primary sources; a vulnerability report or public demonstration is not evidence of a victim incident.

Exploitation evidence source

Technical severity is not a measure of observed attacks or of risk to your deployment. Attempts, demonstrations, and successful compromise retain the qualifications in the source.

How the weakness works

Base64 SVG content can retain embedded script tags through the assistant rendering pipeline.

Conditions required

An authenticated low-privilege caller able to submit the content and active interaction by a user who renders it, as specified by INCIBE’s CVSS vector.

Potential impact

JavaScript execution in the assistant’s browser context and modification of displayed content, with potential confidentiality impact. No affected-customer incident is established.

Affected versions and configurations

The CNA describes the SaaS platform before October 27, 2026 using a date-style range; no semantic release versions are supplied.

Fix and mitigation status

Unknown; INCIBE explicitly says no solution has been reported. The October 27 range boundary must not be treated as proof of a fix.

Practical next steps

Request confirmed remediation from Sisense. As an interim measure, restrict SVG attachment/rendering in AI sessions where possible and avoid rendering untrusted content.

Operating lesson

AI-generated and user-supplied rich content require the same browser sanitization boundaries as other untrusted web content.

Evidence and limits

INCIBE and its CNA record were published October 27. Its exploitation field says No at publication; this is recorded as not established from reviewed sources. The same bulletin’s non-AI BloX vulnerability is separate.

Sources