{"id":"CVE-2026-44721","published":"2026-05-15T20:02:16Z","modified":"2026-05-19T12:37:57Z","title":"Open WebUI: Stored XSS via Model Description","vendor":"open-webui","product":"open-webui","tier":"tail","description":{"text":"Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.0, a stored cross-site scripting (XSS) vulnerability that allows any authenticated user with model creation permission (workspace.models) to execute arbitrary JavaScript in the browser of any other user (including admins) who views the malicious model in the chat UI. This vulnerability is fixed in 0.9.0.","provenance":"source","from":"CVE.org"},"cvss":{"score":7.3,"severity":"HIGH","version":"3.1","vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:R/S:U/C:H/I:H/A:N","provenance":"source","from":"CVE.org CNA"},"epss":{"score":0.00308,"percentile":0.24182,"provenance":"source"},"cwes":[{"id":"CWE-79","name":"Improper Neutralization of Input During Web Page Generation"}],"references":[{"url":"https://github.com/open-webui/open-webui/security/advisories/GHSA-gf5m-wcrh-7928","tags":["x_refsource_CONFIRM"]}],"priority":{"score":9.9,"factors":[{"name":"EPSS percentile","points":4.84},{"name":"CVSS severity","points":5.11}]},"vendorSlug":"open-webui","affectedVersions":[{"vendor":"open-webui","product":"open-webui","versions":["< 0.9.0"]}],"ssvc":{"exploitation":"poc","automatable":"no","technicalImpact":"total","timestamp":"2026-05-19T12:37:29.506330Z","from":"CISA ADP","provenance":"source"},"affected":[{"ecosystem":"npm","package":"open-webui","vulnerableRanges":["<0.9.0"],"patched":["0.9.0"],"from":["GHSA-gf5m-wcrh-7928"],"provenance":"source"},{"ecosystem":"PyPI","package":"open-webui","vulnerableRanges":["<0.9.0"],"patched":["0.9.0"],"from":["GHSA-gf5m-wcrh-7928","PYSEC-2026-2725"],"provenance":"source"}],"summary":{"text":"Open WebUI, the self-hosted AI platform, contains a stored cross-site scripting flaw in how model descriptions are handled, affecting versions before 0.9.0. The flaw is in the model description field, which is rendered unsafely in the chat interface. An attacker needs an account holding model creation permission (workspace.models) and network access to the instance; they can plant a malicious model description that executes arbitrary JavaScript in the browser of any other user who views it in the chat UI, including administrators, and this requires that victim to interact by viewing the model.","provenance":"ai","model":"claude-haiku-5-5","generatedAt":"2026-11-20T22:02:40Z"},"euvd":{"id":"EUVD-2026-30625","exploited":false,"provenance":"source"},"lineage":{"packageHistory":[{"cve":"CVE-2025-64496","published":"2025-11-08","fixed":"0.6.35","kev":false},{"cve":"CVE-2025-65959","published":"2025-12-04","fixed":"0.6.37","kev":false},{"cve":"CVE-2026-45665","published":"2026-05-15","fixed":"0.8.0","kev":false},{"cve":"CVE-2026-44721","published":"2026-05-15","fixed":"0.9.0","kev":false},{"cve":"CVE-2026-45395","published":"2026-05-15","fixed":"0.9.5","kev":false,"link":{"from":"CVE-2026-44721","to":"CVE-2026-45395","relation":"successor","confidence":"low","kind":"sequence","orderedBy":"description","evidence":[{"kind":"version-step","provenance":"source","package":"npm:open-webui","fromFixed":"0.9.0","toFixed":"0.9.5","daysApart":0}]}}],"package":"npm:open-webui"}}