GHSA-8JJP-R2W2-4V22

Vulnerability from github – Published: 2026-05-14 20:26 – Updated: 2026-05-15 23:54
VLAI
Summary
Open WebUI: Low-privilege authenticated users can enumerate and stop global background tasks, causing system-wide chat disruption
Details

Summary

Any authenticated user with low privileges can enumerate active background tasks across the system and stop tasks belonging to other users via the GET /api/tasks and POST /api/tasks/stop/{task_id} methods. This allows a casual user to disrupt system-wide chat usage by continuously canceling other users' active tasks. This is a real authorization vulnerability affecting integrity and usability in multi-user deployments.

Details

Open WebUI exposes GET /api/tasks and POST /api/tasks/stop/{task_id} to any verified user. These endpoints operate on a global task namespace and accept raw task_id values without checking whether the task belongs to the current caller.

As a result, a normal authenticated user can enumerate active global task IDs and stop tasks belonging to other users.

Root cause:

  1. Route authorization is too weak.

In backend/open_webui/main.py, both endpoints only require get_verified_user:

@app.post('/api/tasks/stop/{task_id}')
async def stop_task_endpoint(request: Request, task_id: str, user=Depends(get_verified_user)):
    result = await stop_task(request.app.state.redis, task_id)

@app.get('/api/tasks')
async def list_tasks_endpoint(request: Request, user=Depends(get_verified_user)):
    return {'tasks': await list_tasks(request.app.state.redis)}

get_verified_user accepts both user and admin roles in backend/open_webui/utils/auth.py.

  1. The helper operates on a global namespace.

In backend/open_webui/tasks.py, task listing is global:

async def list_tasks(redis):
    if redis:
        return await redis_list_tasks(redis)
    return list(tasks.keys())

In backend/open_webui/tasks.py, task stopping is by raw task_id:

async def stop_task(redis, task_id: str):
    if redis:
        item_id = await redis.hget(REDIS_TASKS_KEY, task_id)
        await redis_send_command(redis, {'action': 'stop', 'task_id': task_id})
        await redis_cleanup_task(redis, task_id, item_id or None)

There is no owner check, no user_id check, and no mapping from task_id back to the current caller before stop or cleanup.

This also appears unintended because the codebase already has a scoped route, GET /api/tasks/chat/{chat_id}, which checks whether the chat belongs to the current user before returning task IDs.

Relevant code references: - backend/open_webui/main.py:1975 - backend/open_webui/main.py:1984 - backend/open_webui/main.py:1989 - backend/open_webui/tasks.py:127 - backend/open_webui/tasks.py:145 - backend/open_webui/utils/auth.py:415

Suggested remediation: - Store task ownership metadata such as user_id and chat_id, then enforce owner-only access for non-admin users - Suggested implementation locations: - backend/open_webui/main.py: add authentication checks for /api/tasks and /api/tasks/stop/{task_id} - backend/open_webui/tasks.py: add support for storing/querying task ownership metadata such as user_id and chat_id, and support owner-scoped listing/stopping

PoC

Preconditions:

  • Default main branch deployment
  • Authentication enabled
  • Two normal user accounts, or any multi-user deployment where the attacker has one authenticated non-admin account
  • At least one task actively running for another user

This does not require any weakened security settings.

PoC objective:

  1. Show that a non-admin user can see global active task IDs that are not their own
  2. Show that the same user can stop another user's active task

Reproduction steps:

Step 1. Victim starts a long-running task

Using the UI, User A starts a long response generation or another background task and leaves it running.

Expected security model: User B should not be able to see or control User A's task.

Step 2. Attacker enumerates global task IDs

Using User B's authenticated token:

curl -i -H "Authorization: Bearer <USER_B_TOKEN>" http://<open-webui-host>/api/tasks

Expected result:

  • only User B's own task IDs should be returned, or
  • the endpoint should be admin-only

Actual result: the response returns the global active task list.

Example response shape:

{"tasks":["<task-id-a>","<task-id-b>"]}

This exposes task IDs belonging to other users.

Step 3. Attacker stops a foreign task

Pick a task ID that belongs to User A and send:

curl -i -X POST -H "Authorization: Bearer <USER_B_TOKEN>" http://<open-webui-host>/api/tasks/stop/<FOREIGN_TASK_ID>

Expected result:

  • 403 Forbidden, or
  • 404 Not Found for non-owned tasks, or
  • admin-only access

Actual result: the server accepts the request and attempts to stop the foreign task.

Example response shape:

{"status":true,"message":"Task <FOREIGN_TASK_ID> stopped."}

Step 4. Observe boundary violation

User A's running task is interrupted or disappears from the active task set even though User B does not own it.

What actions become possible that should not be possible:

  • enumerate globally active task IDs across users
  • cancel another user's in-progress generation or background work
  • repeat this for every returned task ID, causing broad cross-user disruption

Copy-paste PoC summary:

  1. Enumerate all active tasks as a normal non-admin user
curl -s -H "Authorization: Bearer <USER_B_TOKEN>" http://<open-webui-host>/api/tasks
  1. Stop a task that does not belong to that user
curl -s -X POST -H "Authorization: Bearer <USER_B_TOKEN>" http://<open-webui-host>/api/tasks/stop/<FOREIGN_TASK_ID>

Impact

Type of vulnerability: broken object-level authorization affecting a global runtime control-plane endpoint.

Who is impacted:

  • all users in a multi-user Open WebUI deployment
  • any user currently running a background task, especially chat generation tasks
  • administrators indirectly, because normal users can disrupt system-wide usage without admin privileges

Direct impact:

  • cross-user task ID disclosure
  • cross-user task cancellation

Practical impact:

  • interruption of long-running chat responses
  • interruption of background indexing or ingestion tasks associated with shared runtime jobs
  • one ordinary authenticated low-privilege user can continuously poll /api/tasks and immediately cancel every newly created active task
  • with a simple loop or script, this becomes a practical persistent denial-of-service against chat usage for all users on the instance
  • in a multi-user deployment, normal users may be unable to complete any chat generation while the attacker continues polling and cancelling tasks

Why severity is meaningful:

  • privileges required: low, only an authenticated non-admin account
  • scope: cross-user
  • impact class: integrity and availability
  • exploitation complexity: low once logged in

This is not full account takeover or privilege escalation, but it enables platform-wide operational disruption from a low-privilege account. In practice, sustained exploitation can make chat functionality effectively unusable for other users on the system.

Resolution

Fixed in commit e7ff4768f (#23454, "Add ownership checks to global task endpoints"), first released in v0.9.0 (Apr 2026).

The fix takes a simpler approach than per-task ownership tracking, which would have required a schema change to attribute every task to a user_id:

  • GET /api/tasks and POST /api/tasks/stop/{task_id} are restricted to admin-only via Depends(get_admin_user). Cross-user enumeration and termination are no longer reachable from a non-admin account.
  • A new scoped POST /api/tasks/chat/{chat_id}/stop endpoint covers the legitimate non-admin use case (a user stopping their own in-progress generation), reusing the same chat-ownership check the existing GET /api/tasks/chat/{chat_id} already enforces.

CVE-2025-63681 was a prior disclosure of the same authorization gap against v0.6.33; the fix in v0.9.0 also resolves that.

Users on >= 0.9.0 are not affected.

Show details on source website

{
  "affected": [
    {
      "database_specific": {
        "last_known_affected_version_range": "\u003c= 0.8.12"
      },
      "package": {
        "ecosystem": "PyPI",
        "name": "open-webui"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "0"
            },
            {
              "fixed": "0.9.0"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    }
  ],
  "aliases": [
    "CVE-2026-45399"
  ],
  "database_specific": {
    "cwe_ids": [
      "CWE-862"
    ],
    "github_reviewed": true,
    "github_reviewed_at": "2026-05-14T20:26:49Z",
    "nvd_published_at": "2026-05-15T20:16:48Z",
    "severity": "HIGH"
  },
  "details": "### Summary\nAny authenticated user with low privileges can enumerate active background tasks across the system and stop tasks belonging to other users via the GET /api/tasks and POST /api/tasks/stop/{task_id} methods. This allows a casual user to disrupt system-wide chat usage by continuously canceling other users\u0027 active tasks. This is a real authorization vulnerability affecting integrity and usability in multi-user deployments.\n\n\n### Details\nOpen WebUI exposes `GET /api/tasks` and `POST /api/tasks/stop/{task_id}` to any verified user. These endpoints operate on a global task namespace and accept raw `task_id` values without checking whether the task belongs to the current caller.\n\nAs a result, a normal authenticated user can enumerate active global task IDs and stop tasks belonging to other users.\n\nRoot cause:\n\n1. Route authorization is too weak.\n\nIn `backend/open_webui/main.py`, both endpoints only require `get_verified_user`:\n\n```python\n@app.post(\u0027/api/tasks/stop/{task_id}\u0027)\nasync def stop_task_endpoint(request: Request, task_id: str, user=Depends(get_verified_user)):\n    result = await stop_task(request.app.state.redis, task_id)\n\n@app.get(\u0027/api/tasks\u0027)\nasync def list_tasks_endpoint(request: Request, user=Depends(get_verified_user)):\n    return {\u0027tasks\u0027: await list_tasks(request.app.state.redis)}\n```\n\n`get_verified_user` accepts both `user` and `admin` roles in `backend/open_webui/utils/auth.py`.\n\n2. The helper operates on a global namespace.\n\nIn `backend/open_webui/tasks.py`, task listing is global:\n\n```python\nasync def list_tasks(redis):\n    if redis:\n        return await redis_list_tasks(redis)\n    return list(tasks.keys())\n```\n\nIn `backend/open_webui/tasks.py`, task stopping is by raw `task_id`:\n\n```python\nasync def stop_task(redis, task_id: str):\n    if redis:\n        item_id = await redis.hget(REDIS_TASKS_KEY, task_id)\n        await redis_send_command(redis, {\u0027action\u0027: \u0027stop\u0027, \u0027task_id\u0027: task_id})\n        await redis_cleanup_task(redis, task_id, item_id or None)\n```\n\nThere is no owner check, no `user_id` check, and no mapping from `task_id` back to the current caller before stop or cleanup.\n\nThis also appears unintended because the codebase already has a scoped route, `GET /api/tasks/chat/{chat_id}`, which checks whether the chat belongs to the current user before returning task IDs.\n\nRelevant code references:\n- `backend/open_webui/main.py:1975`\n- `backend/open_webui/main.py:1984`\n- `backend/open_webui/main.py:1989`\n- `backend/open_webui/tasks.py:127`\n- `backend/open_webui/tasks.py:145`\n- `backend/open_webui/utils/auth.py:415`\n\nSuggested remediation:\n- Store task ownership metadata such as `user_id` and `chat_id`, then enforce owner-only access for non-admin users\n- Suggested implementation locations:\n  - `backend/open_webui/main.py`: add authentication checks for `/api/tasks` and `/api/tasks/stop/{task_id}`\n  - `backend/open_webui/tasks.py`: add support for storing/querying task ownership metadata such as `user_id` and `chat_id`, and support owner-scoped listing/stopping\n\n\n\n### PoC\nPreconditions:\n\n- Default `main` branch deployment\n- Authentication enabled\n- Two normal user accounts, or any multi-user deployment where the attacker has one authenticated non-admin account\n- At least one task actively running for another user\n\nThis does not require any weakened security settings.\n\nPoC objective:\n\n1. Show that a non-admin user can see global active task IDs that are not their own\n2. Show that the same user can stop another user\u0027s active task\n\nReproduction steps:\n\n#### Step 1. Victim starts a long-running task\n\nUsing the UI, User A starts a long response generation or another background task and leaves it running.\n\nExpected security model:\nUser B should not be able to see or control User A\u0027s task.\n\n#### Step 2. Attacker enumerates global task IDs\n\nUsing User B\u0027s authenticated token:\n\n```bash\ncurl -i -H \"Authorization: Bearer \u003cUSER_B_TOKEN\u003e\" http://\u003copen-webui-host\u003e/api/tasks\n```\n\nExpected result:\n\n- only User B\u0027s own task IDs should be returned, or\n- the endpoint should be admin-only\n\nActual result:\nthe response returns the global active task list.\n\nExample response shape:\n\n```json\n{\"tasks\":[\"\u003ctask-id-a\u003e\",\"\u003ctask-id-b\u003e\"]}\n```\n\nThis exposes task IDs belonging to other users.\n\n#### Step 3. Attacker stops a foreign task\n\nPick a task ID that belongs to User A and send:\n\n```bash\ncurl -i -X POST -H \"Authorization: Bearer \u003cUSER_B_TOKEN\u003e\" http://\u003copen-webui-host\u003e/api/tasks/stop/\u003cFOREIGN_TASK_ID\u003e\n```\n\nExpected result:\n\n- `403 Forbidden`, or\n- `404 Not Found` for non-owned tasks, or\n- admin-only access\n\nActual result:\nthe server accepts the request and attempts to stop the foreign task.\n\nExample response shape:\n\n```json\n{\"status\":true,\"message\":\"Task \u003cFOREIGN_TASK_ID\u003e stopped.\"}\n```\n\n#### Step 4. Observe boundary violation\n\nUser A\u0027s running task is interrupted or disappears from the active task set even though User B does not own it.\n\nWhat actions become possible that should not be possible:\n\n- enumerate globally active task IDs across users\n- cancel another user\u0027s in-progress generation or background work\n- repeat this for every returned task ID, causing broad cross-user disruption\n\nCopy-paste PoC summary:\n\n1. Enumerate all active tasks as a normal non-admin user\n\n```bash\ncurl -s -H \"Authorization: Bearer \u003cUSER_B_TOKEN\u003e\" http://\u003copen-webui-host\u003e/api/tasks\n```\n\n2. Stop a task that does not belong to that user\n\n```bash\ncurl -s -X POST -H \"Authorization: Bearer \u003cUSER_B_TOKEN\u003e\" http://\u003copen-webui-host\u003e/api/tasks/stop/\u003cFOREIGN_TASK_ID\u003e\n```\n\n### Impact\nType of vulnerability:\nbroken object-level authorization affecting a global runtime control-plane endpoint.\n\nWho is impacted:\n\n- all users in a multi-user Open WebUI deployment\n- any user currently running a background task, especially chat generation tasks\n- administrators indirectly, because normal users can disrupt system-wide usage without admin privileges\n\nDirect impact:\n\n- cross-user task ID disclosure\n- cross-user task cancellation\n\nPractical impact:\n\n- interruption of long-running chat responses\n- interruption of background indexing or ingestion tasks associated with shared runtime jobs\n- one ordinary authenticated low-privilege user can continuously poll `/api/tasks` and immediately cancel every newly created active task\n- with a simple loop or script, this becomes a practical persistent denial-of-service against chat usage for all users on the instance\n- in a multi-user deployment, normal users may be unable to complete any chat generation while the attacker continues polling and cancelling tasks\n\nWhy severity is meaningful:\n\n- privileges required: low, only an authenticated non-admin account\n- scope: cross-user\n- impact class: integrity and availability\n- exploitation complexity: low once logged in\n\nThis is not full account takeover or privilege escalation, but it enables platform-wide operational disruption from a low-privilege account. In practice, sustained exploitation can make chat functionality effectively unusable for other users on the system.\n\n## Resolution\n\nFixed in commit [e7ff4768f](https://github.com/open-webui/open-webui/commit/e7ff4768f8ffe1924b4576381c9e45e8a64350e4) ([#23454](https://github.com/open-webui/open-webui/pull/23454), \"Add ownership checks to global task endpoints\"), first released in **v0.9.0** (Apr 2026).\n\nThe fix takes a simpler approach than per-task ownership tracking, which would have required a schema change to attribute every task to a `user_id`:\n\n- `GET /api/tasks` and `POST /api/tasks/stop/{task_id}` are restricted to admin-only via `Depends(get_admin_user)`. Cross-user enumeration and termination are no longer reachable from a non-admin account.\n- A new scoped `POST /api/tasks/chat/{chat_id}/stop` endpoint covers the legitimate non-admin use case (a user stopping their own in-progress generation), reusing the same chat-ownership check the existing `GET /api/tasks/chat/{chat_id}` already enforces.\n\nCVE-2025-63681 was a prior disclosure of the same authorization gap against v0.6.33; the fix in v0.9.0 also resolves that.\n\nUsers on `\u003e= 0.9.0` are not affected.",
  "id": "GHSA-8jjp-r2w2-4v22",
  "modified": "2026-05-15T23:54:50Z",
  "published": "2026-05-14T20:26:49Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://github.com/open-webui/open-webui/security/advisories/GHSA-8jjp-r2w2-4v22"
    },
    {
      "type": "ADVISORY",
      "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-45399"
    },
    {
      "type": "WEB",
      "url": "https://github.com/open-webui/open-webui/pull/23454"
    },
    {
      "type": "WEB",
      "url": "https://github.com/open-webui/open-webui/commit/e7ff4768f8ffe1924b4576381c9e45e8a64350e4"
    },
    {
      "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:L/UI:N/S:U/C:L/I:N/A:H",
      "type": "CVSS_V3"
    }
  ],
  "summary": "Open WebUI: Low-privilege authenticated users can enumerate and stop global background tasks, causing system-wide chat disruption"
}


Log in or create an account to share your comment.




Tags
Taxonomy of the tags.


Loading…

Loading…

Loading…

Forecast uses a logistic model when the trend is rising, or an exponential decay model when the trend is falling. Fitted via linearized least squares.

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.

Loading…

Detection rules are retrieved from Rulezet.

Loading…

Loading…