Hackers exploit a critical Flowise flaw affecting thousands of AI workflows

A critical vulnerability in the Flowise low-code platform has been exploited by hackers, allowing them to inject malicious JavaScript code into thousands of AI workflows. The flaw, rated as max-severity, is attributed to a design oversight in the platform's custom MCP node, which enables connectivity between AI agents and external tools. As a result, nearly 15,000 Flowise instances have been exposed, with threat actors already leveraging the vulnerability to insert arbitrary JavaScript code¹. The exploitation of this flaw poses significant risks to organizations relying on Flowise for building custom large language models and agentic systems. The vulnerability highlights the importance of securing low-code platforms, which are increasingly being used to develop and deploy AI-powered applications. This vulnerability matters to practitioners because it underscores the need for rigorous security testing and validation of low-code platforms to prevent similar exploits.