A severe remote code execution (RCE) vulnerability in the AI low-code platform Langflow is being exploited by threat actors to harvest credentials and gain root access.
- CVE-2026-0768 allows unauthenticated attackers to execute arbitrary Python code as root.
- All Langflow versions up to 1.4.2 are susceptible to this critical flaw.
- Increased targeting from Russian-origin queries focusing on SSH access and secret keys.
Threat actors have begun aggressively exploiting a critical-severity remote code execution (RCE) vulnerability within Langflow, a popular AI low-code platform. According to vulnerability intelligence firm VulnCheck, the flaw, tracked as CVE-2026-0768, carries a devastating CVSS score of 9.8, indicating an extreme level of risk.
The vulnerability resides in the code validator of Langflow’s custom component editor. Because user-supplied strings are not properly sanitized or validated before being passed for Python execution, an attacker can bypass authentication entirely and execute arbitrary code with root privileges on the host system.
Why This Matters
BozokMedia analysis shows that the shift toward AI-driven automation is creating a new, vast attack surface. Langflow serves as a bridge between complex LLMs and operational workflows; therefore, a compromise here doesn't just leak data—it potentially gives attackers a programmable gateway into the heart of an organization's AI infrastructure.
The rapid adoption of AI low-code tools often outpaces the implementation of rigorous security audits, leaving a window of opportunity for sophisticated state-sponsored actors.
VulnCheck has observed a surge in reconnaissance activities, with a significant number of queries originating from Russia. These attackers are specifically hunting for environment variables, secret keys, and SSH access. In the UK alone, over 360 exploitation attempts were detected hitting canary systems by Monday.
This is not an isolated incident. While 2025 saw minimal exploitation of Langflow, 2026 has witnessed a dramatic spike. VulnCheck reports that 11 additional vulnerabilities have been targeted this year, with over 15,000 successful attacks recorded across three specific flaws.
| CVE ID | Severity | Primary Impact |
|---|---|---|
| CVE-2026-0768 | Critical (9.8) | Unauthenticated Root RCE |
| CVE-2026-0769 | High | Unauthorized Code Execution |
| CVE-2025-3248 | Medium/High | Credential Theft |
The security community warns that 'silent patches'—updates released without full disclosure—are not a solution. Instead, they often provide a roadmap for attackers to reverse-engineer the flaw while leaving defenders in the dark about the actual risk level.
Frequently Asked Questions
Q1: Which versions of Langflow are affected?
All releases up to and including version 1.4.2 are vulnerable. Users are urged to update to the latest patched version immediately.
Q2: How can I tell if my system was breached?
Check your logs for unusual Python execution requests in the component editor and monitor for unauthorized SSH access or changes to environment variables.
