Critical Code Injection Vulnerability Discovered in Langflow

What changed

A significant vulnerability has been uncovered in Langflow, a tool designed for creating AI-driven agents and workflows. Specifically, the issue resides in the POST /api/v1/build_public_tmp/{flow_id}/flow endpoint, which permits the construction of public flows without any authentication requirements.

This flaw allows attackers to inject arbitrary Python code into the flow data, which is then executed without any form of sandboxing, leading to potential unauthorized access and control over the system.

Who is affected

The vulnerability impacts various users and organizations utilizing Langflow, especially those who have not updated to version 1.9.0 or later. The following groups are particularly at risk:

Developers and teams using Langflow for AI workflow automation, organizations deploying Langflow in production environments, and any third-party services relying on Langflow for public flow building.

• Langflow users operating versions prior to 1.9.0
• Organizations leveraging AI-powered workflows
• Developers integrating Langflow into applications

What to do

To mitigate the risk associated with this vulnerability, immediate action is recommended. Users should upgrade to the latest version of Langflow to ensure that they are protected against this exploit.

Additionally, it is advisable to review security practices surrounding the deployment of AI tools and consider implementing additional layers of authentication and validation.

• Upgrade to Langflow version 1.9.0 or later
• Implement strong authentication measures
• Regularly review and audit security configurations