CISA Orders Urgent Patching of Langflow RCE Flaw

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) on Tuesday ordered federal agencies to prioritize fixing a vulnerability in Langflow, a visual framework for building AI agents. The flaw is already being actively exploited by attackers.

What the vulnerability is

The flaw has been assigned the identifier CVE-2026-0770 and rated critical. It lets an unauthenticated attacker achieve remote code execution with root privileges, and the attack is low in complexity — no special conditions are required.

According to the Trend Micro researchers who discovered it, the vulnerability lies in the handling of the exec_globals parameter passed to the validate endpoint. Because a resource is pulled in from an untrusted control scope, an attacker is able to execute arbitrary code in the root context.

Exploitation in the wild

KEVIntel, a company specializing in vulnerability intelligence, first observed exploitation of CVE-2026-0770 on June 27. Before CISA added the flaw to its Known Exploited Vulnerabilities (KEV) catalog, more than 220 attack attempts had been logged from 64 unique IP addresses.

KEVIntel founder Ryan Dewhurst told BleepingComputer that this went beyond mere vulnerability scanning: attempts were observed to deploy malware and steal AWS credentials, environment variables and container metadata. Most requests amounted to command execution and system reconnaissance, but downloads of second-stage scripts and access to files containing secrets were also recorded.

For organizations using Langflow, the expert advises reviewing the request history for /api/v1/validate/code, checking activity on the hosts, restricting access to the validation function, and rotating credentials wherever successful code execution cannot be ruled out.

The deadline is Friday

By adding the vulnerability to the KEV catalog, CISA has required civilian federal agencies to close it by Friday, in accordance with the binding directive BOD 26-04. The agency noted that flaws of this kind regularly become an attack vector and pose serious risks to government infrastructure.

This is far from the first Langflow vulnerability to come to CISA's attention. The agency previously flagged exploitation of a missing-authentication issue (CVE-2025-3248) in May 2025, a code injection vulnerability (CVE-2026-33017) in March 2026, and an IDOR flaw (CVE-2026-55255) in early July. What's more, CVE-2025-3248 is already being used in ransomware attacks: according to Sysdig, the JadePuffer group is leveraging it to exfiltrate the contents of the PostgreSQL databases used by Langflow.

Source: Bleeping Computer