A newly disclosed zero-day vulnerability, CVE-2025-3248, is being actively exploited in the wild—and it’s targeting Langflow, a popular open-source framework for building AI-powered applications. In this episode, we unpack how a missing authentication check in the /api/v1/validate/code endpoint allowed remote attackers to run arbitrary code on unpatched servers. With a critical CVSS score of 9.8 and confirmation from CISA’s Known Exploited Vulnerabilities Catalog, this flaw has serious implications for organizations using versions prior to 1.3.0.
We explore the technical mechanics behind the exploit—including abuse of Python decorators and default arguments—and highlight evidence of real-world attacks detected by honeypots and TOR-sourced payloads. Whether you’re running Langflow or managing open-source AI tools, this is a wake-up call for patching, hardening, and reassessing how you expose development platforms to the internet.
Stay ahead of the threat. Tune in now to learn what went wrong, what’s being done, and what you can do to protect your infrastructure.
