Open WebUI: information disclosure
Medium4.3CVE-2026-45387 · Published May 14, 2026 · updated Jul 13, 2026
### Summary When setting model permissions so that a group has read access to it, intending for other users to use it, those users also can read the model's system prompt. However users may consider their system prompt confidential, so we consider this a security issue. Compare https://genai.owasp.org/llmrisk/llm072025-system-prompt-leakage/ or prompt injections to get popular chatbots on the internet to reveal their prompt. ### Details We discovered that users can open the workspace model edit page /workspace/models/edit?id=notmymodel for models that do not appear in their workspace. Saving is not possible, that permission check is correct. On the API level: - ```/api/v1/models/model?id=notmymodel``` -> returns the model details, most importantly params.system - even though ```/api/v1/models/list``` does NOT contain the model since it checks for write permission. - ```/api/models``` contains the model correctly and does not reveal the system prompt. It seems inconsistent that the REST API list does not contain an item, but if you know the id, you can access it anyway. ### PoC - create model - give read permission to group with another user - other user can access ```/api/...
Affected versions
| Package | Affected | Fixed in |
|---|---|---|
| open-webui PyPI | < 0.9.5 | 0.9.5 |
Details and references
### Summary When setting model permissions so that a group has read access to it, intending for other users to use it, those users also can read the model's system prompt. However users may consider their system prompt confidential, so we consider this a security issue. Compare https://genai.owasp.org/llmrisk/llm072025-system-prompt-leakage/ or prompt injections to get popular chatbots on the internet to reveal their prompt. ### Details We discovered that users can open the workspace model edit page /workspace/models/edit?id=notmymodel for models that do not appear in their workspace. Saving is not possible, that permission check is correct. On the API level: - ```/api/v1/models/model?id=notmymodel``` -> returns the model details, most importantly params.system - even though ```/api/v1/models/list``` does NOT contain the model since it checks for write permission. - ```/api/models``` contains the model correctly and does not reveal the system prompt. It seems inconsistent that the REST API list does not contain an item, but if you know the id, you can access it anyway. ### PoC - create model - give read permission to group with another user - other user can access ```/api/v1/models/model?id=notmymodel``` ### Impact System prommpt leakage If this is intended behavior for the "read" permission, maybe there should be an additional "use" permission (which would be 99% of use cases of the read permission i believe).
- CVSS 3.1
- CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:N/A:N
- Severity from
- GitHub (reviewed advisory)
- Weakness
- CWE-200
- Also known as
- CVE-2026-45387, PYSEC-2026-2728
More Open WebUI advisories
All Open WebUI| Date | Advisory | Severity | Fixed in |
|---|---|---|---|
| May 14 | Open WebUI: LDAP and OAuth First-User Race Condition Allows Multiple Admin Accounts High8.1May 14 | High8.1 | 0.9.0 |
| May 14 | Open WebUI: Jupyter code execution works despite `ENABLE_CODE_EXECUTION=false` , feature gate bypassedHigh8.8May 14 | High8.8 | 0.8.12 |
| May 14 | Open WebUI: shared-chat branch ignores access_type, allowing unauthorized file deletionHigh8.0May 14 | High8.0 | 0.9.0 |
| May 14 | Open WebUI: Unauthenticated endpoint can trigger embedding generation (cost/DoS)Medium6.5May 14 | Medium6.5 | 0.8.0 |
| May 14 | Open WebUI has an Indirect Object Reference (IDOR) in user notesMedium6.5May 14 | Medium6.5 | 0.8.11 |
| May 14 | Open WebUI: insecure direct object referenceHigh8.1May 14 | High8.1 | 0.9.5 |