A Critical Vulnerability in Langflow Framework Exposes OpenAI and AWS Keys to Attackers
Threat actors have been exploiting a critical vulnerability in Langflow, an open-source framework for building AI applications, to steal credentials, tokens, and keys from vulnerable instances. According to threat intelligence firm VulnCheck, over 360 attacks were detected on its honeypots in the UK over the weekend, with most of the traffic originating from Russia.
The vulnerability, identified as CVE-2026-0768, is an unauthenticated remote code execution (RCE) flaw that resides in the code validator of Langflow’s custom component editor. It allows attackers to execute arbitrary code without authentication, granting them root privileges. This critical issue was disclosed in January and affects Langflow versions 1.4.2 and earlier.
Langflow is a Python-based low-code platform used for building AI applications, agents, chatbots, and retrieval-augmented generation systems. Users create workflows by connecting components for language models, prompts, databases, APIs, and other tools through a graphical interface. However, the framework’s reliance on user-supplied input makes it vulnerable to attacks like this one.
Attackers are using this vulnerability to harvest administrative credentials or superuser authentication keys for Langflow instances, AWS secrets, and OpenAI API keys. They accomplish this by conducting reconnaissance and querying environment variables, which allows them to gain access to sensitive data. Once attackers have valid credentials, they can bypass most security measures, as only 37% of their actions are blocked.
This is not the first time Langflow has been targeted by attackers this year. In March, a critical code-injection flaw (CVE-2026-33017) was exploited within a day of its disclosure, allowing attackers to execute Python scripts and harvest sensitive data. Since then, multiple other vulnerabilities have been discovered in the framework.
To mitigate this risk, Langflow users are urged to upgrade to the latest available version, 1.11.6, which addresses all known flaws in the popular tool. This is a critical step in securing AI applications and preventing unauthorized access to sensitive data.
The discovery of CVE-2026-0768 serves as a reminder of the importance of regular security updates and patching. Attackers will continue to exploit vulnerabilities like this one unless users take proactive steps to protect themselves. By staying up-to-date with the latest security patches and best practices, AI developers and administrators can reduce their exposure to these types of attacks and safeguard their applications against unauthorized access.
As a practical takeaway, it’s essential for Langflow users to prioritize upgrading to the latest version of the framework as soon as possible. Additionally, organizations should implement robust monitoring and detection tools to identify potential security threats and respond quickly in case of an attack. By doing so, they can minimize the risk of data breaches and ensure the continued integrity of their AI applications.
Source: Bleeping Computer — 2026-09-01