CISA Confirms Langflow RCE Flaw Exploitation

The U.S. Cybersecurity and Infrastructure Security Agency has confirmed active exploitation of a critical vulnerability in the open-source Langflow framework, a popular tool for building AI agent workflows, and added it to its Known Exploited Vulnerabilities catalog. The flaw, identified as CVE-2026-33017, enables unauthenticated attackers to execute arbitrary code remotely and was first disclosed on March 17, 2026.

Langflow, which has garnered over 145,000 stars on GitHub for its drag-and-drop interface that connects AI nodes into executable pipelines via a REST API, affects versions up to 1.8.1. The vulnerability stems from an exposed API endpoint designed for building public flows without authentication. According to the National Vulnerability Database, this endpoint, known as "build_public_tmp," incorrectly accepts attacker-supplied data containing embedded Python code in flow node definitions. The application executes this code using Python's exec function without any sandboxing, allowing a single crafted HTTP request to compromise affected systems.

Exploitation began rapidly after disclosure. Sysdig researchers reported observing scanning attempts on vulnerable instances...