Attackers exploit critical Langflow RCE within hours as CISA sounds alarm

Summary

Attackers quickly exploited a critical remote code execution (RCE) vulnerability in the open-source AI tool Langflow within 20 hours of its disclosure. The flaw, identified as CVE-2026-33017, allows unauthenticated code injection through malicious workflow data, enabling attackers to run arbitrary Python code on vulnerable instances. CISA has added this vulnerability to its Known Exploited Vulnerabilities (KEV) catalog, mandating urgent remediation for federal agencies.

IFF Assessment

FOE

The rapid exploitation of a critical vulnerability in an AI development tool by threat actors signifies a significant risk to systems utilizing this technology.

Severity

9.8 Critical

The CVSS score of 9.8 reflects the critical nature of the unauthenticated RCE vulnerability. Its high exploitability (attack vector: network, privileges required: none, user interaction: none) and significant impact (confidentiality, integrity, and availability are all high) make it a severe threat.

Defender Context

This incident highlights the growing trend of attackers targeting popular AI development frameworks, potentially leading to widespread compromise of AI infrastructure. Defenders must prioritize patching and securing any deployments of Langflow, and remain vigilant for exploitation attempts on similar AI-related tools.

Read Full Story →