CVE-2025-47783: HumanSignal Label Studio reflected XSS in label_config parameter

CVE-2025-47783 · Severity: high · CVSS 4 · Published 2025-05-15

Label Studio, a popular open-source data labeling platform, is vulnerable to a security flaw that allows attackers to run malicious scripts in a user's browser. By tricking a user into clicking a link or visiting a malicious website, an attacker could perform unauthorized actions on the user's behalf within the application. While session cookies are protected, this could still lead to data exposure or unauthorized modifications to labeling projects.

Technologies: label-studio (PyPI). Vendors: PyPI.

Quick answers

Is CVE-2025-47783 exploited in the wild?
No. Junglewise Threat Intelligence has no report of CVE-2025-47783 being exploited in the wild.
How severe is CVE-2025-47783?
High severity, with a CVSS score of 4, published 15 May 2025.
Which products and versions are affected?
label-studio (PyPI) < 1.18.0, ECOSYSTEM:introduced=0,fixed=1.18.0.
Is there a fix for CVE-2025-47783?
Yes. label-studio (PyPI): upgrade to 1.18.0 or later. Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.
How is CVE-2025-47783 attacked?
It is attacked over the network, in a network appliance.
Am I affected by CVE-2025-47783?
Check whether you run label-studio (PyPI) at an affected version. Junglewise does that check for you around the clock: connect your code, cloud and domains, and it matches new vulnerabilities like this one against what you actually run and tells you when one reaches you.

How to fix

  • label-studio (PyPI): upgrade to 1.18.0 or later
    pip install "label-studio>=1.18.0"

Technical details

A reflected Cross-Site Scripting (XSS) vulnerability exists in Label Studio's `POST /projects/upload-example/` endpoint. The root cause is located in `label_studio/projects/views.py`, where the `label_config` POST parameter is processed and subsequently returned in an `HttpResponse` using `json.dumps()` without sufficient sanitization or proper Content-Type headers to prevent browser execution. An attacker can exploit this by crafting a malicious request (e.g., via a hidden form on a third-party site) that executes arbitrary JavaScript in the victim's browser session. While `HttpOnly` flags on session cookies mitigate direct session hijacking, the attacker can still perform any action the user is authorized to do. The vulnerability is patched in version 1.18.0.

Affected products

  • HumanSignal label-studio < 1.18.0

Identifiers

  • CVE-2025-47783
  • GHSA-8jhr-wpcm-hh4h
  • PYSEC-2025-124

Timeline

  • 2025-05-14: disclosed
  • 2025-05-15: advisory: GitHub Advisory published
  • 1.18.0: patched

References

This page as Markdown. Free to reuse under CC BY 4.0 with a link back.