GHSA-4QG5-CXX4-G927
Vulnerability from github – Published: 2026-09-10 15:09 – Updated: 2026-09-10 15:09Summary
Open WebUI's OAuth token exchange endpoint issues a session for a provider access token without applying the email domain allowlist that the normal OAuth login callback enforces. An account whose email domain the login callback would refuse could still obtain a working session through this endpoint.
Preconditions
ENABLE_OAUTH_TOKEN_EXCHANGE=True. It is disabled by default, so a default deployment is not affected.OAUTH_ALLOWED_DOMAINSset to something other than*. Deployments without a domain allowlist are not affected.- A valid, unexpired access token on the configured provider.
- An Open WebUI account already linked to that provider subject, or an account with a matching email when
OAUTH_MERGE_ACCOUNTS_BY_EMAILis enabled. This endpoint never creates accounts, so a token for a subject with no existing account is rejected.
Impact
An admin who narrows the domain allowlist expects users outside it to lose access at their next sign-in. The login callback does deny them. Token exchange kept issuing sessions, so a user whose domain was removed retained working access as their existing account at its existing role. The endpoint cannot create an account and cannot raise anyone's role, so this grants continued access rather than new or elevated access.
Fix
fb5ef978b, released in 0.9.0, adds the same domain allowlist check to the token exchange endpoint that the login callback runs, and denies the exchange with 403 when the email domain is not allowed. Upgrading restores the check with no further action.
Root cause
The affected component is the OAuth token exchange endpoint in backend/open_webui/routers/auths.py, present in builds from 0.8.0 onward.
The endpoint was added as a second entry point into the same session-issuing path the OAuth login callback uses, but it re-implemented only the identity lookup and not the policy checks surrounding it. The domain allowlist check lived inside the callback's own body rather than in shared code, so the second caller inherited none of it.
Credits
@Classic298
{
"affected": [
{
"package": {
"ecosystem": "PyPI",
"name": "open-webui"
},
"ranges": [
{
"events": [
{
"introduced": "0.8.0"
},
{
"fixed": "0.9.0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-88005"
],
"database_specific": {
"cwe_ids": [
"CWE-863"
],
"github_reviewed": true,
"github_reviewed_at": "2026-09-10T15:09:20Z",
"nvd_published_at": null,
"severity": "MODERATE"
},
"details": "## Summary\nOpen WebUI\u0027s OAuth token exchange endpoint issues a session for a provider access token without applying the email domain allowlist that the normal OAuth login callback enforces. An account whose email domain the login callback would refuse could still obtain a working session through this endpoint.\n\n## Preconditions\n- `ENABLE_OAUTH_TOKEN_EXCHANGE=True`. It is disabled by default, so a default deployment is not affected.\n- `OAUTH_ALLOWED_DOMAINS` set to something other than `*`. Deployments without a domain allowlist are not affected.\n- A valid, unexpired access token on the configured provider.\n- An Open WebUI account already linked to that provider subject, or an account with a matching email when `OAUTH_MERGE_ACCOUNTS_BY_EMAIL` is enabled. This endpoint never creates accounts, so a token for a subject with no existing account is rejected.\n\n## Impact\nAn admin who narrows the domain allowlist expects users outside it to lose access at their next sign-in. The login callback does deny them. Token exchange kept issuing sessions, so a user whose domain was removed retained working access as their existing account at its existing role. The endpoint cannot create an account and cannot raise anyone\u0027s role, so this grants continued access rather than new or elevated access.\n\n## Fix\nfb5ef978b, released in 0.9.0, adds the same domain allowlist check to the token exchange endpoint that the login callback runs, and denies the exchange with 403 when the email domain is not allowed. Upgrading restores the check with no further action.\n\n## Root cause\nThe affected component is the OAuth token exchange endpoint in `backend/open_webui/routers/auths.py`, present in builds from 0.8.0 onward.\n\nThe endpoint was added as a second entry point into the same session-issuing path the OAuth login callback uses, but it re-implemented only the identity lookup and not the policy checks surrounding it. The domain allowlist check lived inside the callback\u0027s own body rather than in shared code, so the second caller inherited none of it.\n\n## Credits\n@Classic298",
"id": "GHSA-4qg5-cxx4-g927",
"modified": "2026-09-10T15:09:20Z",
"published": "2026-09-10T15:09:20Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/open-webui/open-webui/security/advisories/GHSA-4qg5-cxx4-g927"
},
{
"type": "WEB",
"url": "https://github.com/open-webui/open-webui/pull/23639"
},
{
"type": "WEB",
"url": "https://github.com/open-webui/open-webui/commit/fb5ef978bfb451c3f2221931e08e54250cec58ca"
},
{
"type": "PACKAGE",
"url": "https://github.com/open-webui/open-webui"
},
{
"type": "WEB",
"url": "https://github.com/open-webui/open-webui/releases/tag/v0.9.0"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:N",
"type": "CVSS_V3"
}
],
"summary": "Open WebUI: Users denied by the OAuth domain allowlist or role policy can still sign in via token exchange"
}
Drag nodes to arrange them. Double-click a related vulnerability to open it, or a source to visit its link. The Physics button on the left tunes the layout and the simulation; the legend at the bottom left hides or shows nodes by kind and by sighting type.
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
Vulnerability timeline
Sightings and recorded dates, grouped by day · UTC
How durations are calculated
Following the VTEM framework (Black Hat USA 2015), time to patch starts when the vendor or a coordinator was informed, total time to patch at discovery, and disclosure to patch at public disclosure; vendor response time runs from the report to its acknowledgement. Time to exploit and time unpatched start at the earliest discovery or report, falling back to public disclosure. Unpatched time ends at the first recorded patch, or the calculation time when no patch is recorded. Exploitation to patch shows the interval between those two events; a published proof of concept is a public exploit, not exploitation.
Each result shows its source dates. A negative interval means the end event came first; when date-only inputs leave the order open, the result says so. Date-only inputs are approximate. The earliest evidence wins: KEV report dates and sightings are labelled as such. Records describing another vulnerability are listed but never calculated from, and a PoC alone does not establish exploitation. Missing patch evidence does not prove that no patch exists. Filters below change the event list only.
Loading timeline…
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.
Browse all ATT&CK techniques and the vulnerabilities related to each.
Related by attack behaviour
Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.