Critical Flaw in Langflow AI Platform Under Attack

A critical unauthenticated code-injection vulnerability (CVE-2026-33017, CVSS 9.8) in the Langflow AI-agent framework was disclosed and observed being exploited within 24 hours; attackers could supply attacker-controlled flow data that gets executed via exec(), enabling remote code execution, extraction of API keys/credentials, and potential lateral movement. Langflow released version 1.9.0 to mitigate the issue and researchers warn that rapid exploitation of open-source AI tooling is becoming common, urging rapid patching, runtime detection, and network segmentation.