Hackers are actively exploiting a critical vulnerability in Langflow, a widely-used low-code AI development platform, which enables remote code execution. Identified as CVE-2026-5027 with a CVSS score of 8.8, this security flaw involves a path traversal issue that lets attackers write files to any location on the system. The vulnerability is found in the ‘POST /api/v2/files’ endpoint, where the ‘filename’ parameter is not properly sanitized, allowing path traversal sequences like ‘../’. This flaw permits unauthenticated attackers to execute arbitrary code on vulnerable systems, as Langflow’s default setting allows auto-login without credentials. Hackers can send a single unauthenticated request to acquire a valid session token and exploit the vulnerability. Numerous Langflow instances, particularly in North America, are exposed to this risk, highlighting a trend where attackers target AI development infrastructure and tools. The vulnerability was publicly disclosed on March 27 after multiple failed attempts to alert the developers. Langflow has been contacted for a statement, and updates will be provided as they respond.
How are hackers exploiting the Langflow vulnerability for remote code execution?
