/

CVE-2026-45399

Severity:

HIGH RISK

(7.1)

Broken / Missing Authorization in Open WebUI Task

CVE-2026-45399

Analyzed

Vulnerability type

Category:

Broken / Missing Authorization

CWE:

CWE-862

Impact:

Cross-User Task Disruption and Service Interruption

Severity:

HIGH RISK

(7.1)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

Open WebUI contained a missing authorization check in its background task management APIs, allowing any authenticated low-privileged user to view and stop other users' tasks.

In multi-user environments, this flaw can be abused to disrupt ongoing chats and background processing by repeatedly canceling tasks initiated by other users.

Affected context

This issue affects Open WebUI deployments where:

  • Multiple users share a single Open WebUI instance
  • The /api/tasks and /api/tasks/stop/{task_id} endpoints are exposed to all authenticated users
  • Background tasks (such as chat completions or inference requests) are long-running and can be interrupted

The vulnerability is primarily relevant in multi-user setups where users are expected to be isolated from one another's tasks and workflows.

Exploit preconditions

  • Attack surface: HTTP APIs exposed by Open WebUI, specifically:
  • GET /api/tasks
  • POST /api/tasks/stop/{task_id}
  • Required access: Any valid authenticated account, including low-privileged or standard users
  • Trigger point: Sending API requests to list active tasks and issuing stop commands for task IDs owned by other users
  • Why exploitation is possible: The task listing and stopping logic lacks proper server-side authorization checks to ensure that a caller is permitted to act on a given task, enabling cross-user interference.

Why this happens

The root cause is a missing authorization enforcement on sensitive task management operations.

The backend exposes endpoints that:

  • Enumerate active tasks across all users
  • Allow stopping tasks by task identifier

Without verifying that a task belongs to the requesting user or that the caller has an administrative role, these endpoints violate the principle of least privilege.

This maps to CWE-862: Missing Authorization, where an application relies solely on authentication and fails to enforce fine-grained access controls for protected actions.

Potential impact

  • Disruption of other users' chat sessions by canceling their active tasks
  • Interruption of long-running inference or background processing jobs
  • Degradation of overall system reliability and user trust in the platform
  • Potential denial-of-service-like behavior in shared environments, as one user may repeatedly stop others' tasks

General mitigation guidance

Effective mitigation focuses on adding fine-grained authorization and isolating user operations:

  • Enforce per-user ownership checks on task resources so that users can only view and control their own tasks, unless explicitly granted administrative privileges.
  • Introduce role-based access control (RBAC) for task administration, restricting global task enumeration and termination to trusted roles.
  • Design APIs with resource scoping (e.g., user-scoped queries) so that multi-tenant behavior is explicit and can be centrally enforced.
  • Apply the principle of least privilege in API design, ensuring that default user roles have only the minimum task control capabilities required for normal operation.

How teams detect issues like this

Organizations can identify similar authorization flaws through:

  • Static analysis and code review focused on:
  • Endpoints that list or control shared resources (tasks, jobs, sessions)
  • Missing checks that tie resources to a specific user or tenant
  • Authorization-focused tests that:
  • Attempt to access or manipulate resources created by other accounts
  • Validate that non-admin users cannot perform administrative task actions
  • Policy enforcement in CI/CD pipelines that require explicit authorization annotations or decorators on sensitive routes.

Automated security testing tools or API gateways with centralized authorization policies can help surface and prevent these patterns.

Sources

SourceLink
github.comgithub.com/open-webui/open-webui/security/advisories/GHSA-8jjp-r2w2-4v22

Frequently asked

questions

Yes, in the sense that any authenticated user who can reach the Open WebUI API can exploit it by calling the task listing and stopping endpoints, regardless of their privilege level.

The attacker needs only a valid, low-privileged account on the Open WebUI instance.

No administrative permissions are required, and there is no need for direct server or database access.

Implement strict server-side authorization checks on all task-related endpoints.

Concretely, ensure that each task operation verifies:

  • The identity of the caller
  • The ownership or permitted scope of the task
  • The caller's role or permission where cross-user actions are intended (e.g., admin-only operations)

Adopt a consistent authorization framework that:

  • Centralizes access-control decisions (e.g., middleware, decorators, or policy engines)
  • Treats every endpoint that accesses shared resources as sensitive by default
  • Enforces tenant and user isolation with explicit ownership checks
  • Includes authorization-specific test cases in automated test suites to catch cross-user access paths early.
Find out what's already
exploitable in your codebase.