Executive brief
IBM Langflow, a visual framework for building multi-agent AI applications, contains a critical vulnerability that allows unauthorized individuals to take complete control of the server. By exploiting a flaw in how the system handles logins and code validation, an attacker can remotely execute commands without needing a username or password. This could lead to the theft of sensitive AI models, data exposure, or a total shutdown of the AI development environment.
Technical details
A critical remote code execution (RCE) vulnerability exists in IBM Langflow OSS versions 1.0.0 through 1.10.0 due to an insecure API chain. An unauthenticated attacker can first call the `/api/v1/auto_login` endpoint, which incorrectly grants SUPERUSER tokens to any network requester. The attacker can then use this elevated privilege to access the `/api/v1/validate/code` endpoint, which utilizes the Python `exec()` function to process user-supplied input. This combination allows for arbitrary code execution on the underlying host. The vulnerability is present in default deployments and has been reported as exploited in the wild.
Affected products
- IBM Langflow OSS 1.0.0 through 1.10.0
Timeline
- 2026-07-17: disclosed: Initial report from IBM Corporation
- 2026-08-04: advisory: NVD publication date
- 2026-08-04: exploited: Reported as exploited in the wild in advisory summary