Executive brief
This report concerned Open WebUI, an open-source web interface used to interact with AI chat models, and alleged that login credentials could be exposed in plaintext to an attacker on the same network. However, after review, the issue was determined not to be a genuine security flaw and the advisory was formally withdrawn. No action is required, and organizations using Open WebUI should not treat this as a confirmed risk.
Technical details
The original report (ZDI-CAN-28259 / CVE-2026-0767) described a Cleartext Transmission of Sensitive Information (CWE-319) issue in Open WebUI (<= 0.6.33), claiming that credentials sent to an unspecified endpoint were transmitted in plaintext, allowing an adjacent-network, unauthenticated attacker to sniff traffic and capture credentials for further compromise. The reported CVSS v3.0 vector was AV:A/AC:H/PR:N/UI:N/S:U/C:H/I:N/A:N (score 5.3). GitHub Security Advisories subsequently reviewed and withdrew this advisory, indicating it does not describe a valid vulnerability — likely because standard TLS/HTTPS deployment mitigates the described scenario or the finding was not reproducible/applicable as a distinct flaw. No patched version was ever published (first_patched_version: null), consistent with the advisory being invalid rather than an unpatched issue. Security teams should disregard this CVE as an actionable finding.
Affected products
- Open WebUI open-webui <= 0.6.33
Timeline
- 2026-01-23: disclosed: Advisory published to GitHub Advisory Database / NVD
- 2026-09-02: other: Advisory reviewed and withdrawn by GitHub as not a valid vulnerability