Censys Finds 63% Jump in Exposed AI Infrastructure
The number of AI-related systems accessible on the public internet increased by 63% over a nine-month period, according to data from Censys. The research firm shared these findings first with Axios, highlighting a rapid expansion of digital assets that require security management. Censys identified nearly 268,000 internet-facing systems running 44 distinct AI-related products as of July. This figure represents a significant rise from the approximately 164,000 systems detected in October 2025. The exposed software includes tools that link AI models to corporate databases and internal applications. Emily Austin, a principal researcher at Censys and co-author of the report, noted that these tools often connect to highly sensitive systems. She explained that the risk extends beyond the AI tool itself to the broader network of services it accesses. Zakir Durumeric, CEO of Censys, stated that companies are already struggling to track their existing software inventory. He added that the ease of building and hosting AI applications has made it more difficult for IT teams to monitor what employees are using. These new applications can frequently connect to sensitive company information, including emails, financial records, and internal databases. The most widely exposed product identified was n8n, a workflow automation platform that connects AI agents to other services. Censys found nearly 114,000 internet-facing deployments of n8n. Of these, roughly 77,000 were running software versions affected by at least one recently disclosed critical vulnerability. Researchers also detected over 54,000 instances of OpenClaw control panels left open to the web. This open-source agent allows users to link their personal accounts and various other services to the platform. The data does not confirm whether these specific deployments are actively vulnerable to exploitation. However, the findings highlight the risks associated with exposing software that has access to sensitive company systems. Austin advised that companies need to treat AI infrastructure like any other deployed software. This includes regularly reviewing, updating, and securing these systems to mitigate potential threats.
The number of AI-related systems accessible on the public internet increased by 63% over a nine-month period, according to data from Censys. The research firm shared these findings first with Axios, highlighting a rapid expansion of digital assets that require security management. Censys identified nearly 268,000 internet-facing systems running 44 distinct AI-related products as of July. This figure represents a significant rise from the approximately 164,000 systems detected in October 2025. The exposed software includes tools that link AI models to corporate databases and internal applications. Emily Austin, a principal researcher at Censys and co-author of the report, noted that these tools often connect to highly sensitive systems. She explained that the risk extends beyond the AI tool itself to the broader network of services it accesses. Zakir Durumeric, CEO of Censys, stated that companies are already struggling to track their existing software inventory. He added that the ease of building and hosting AI applications has made it more difficult for IT teams to monitor what employees are using. These new applications can frequently connect to sensitive company information, including emails, financial records, and internal databases. The most widely exposed product identified was n8n, a workflow automation platform that connects AI agents to other services. Censys found nearly 114,000 internet-facing deployments of n8n. Of these, roughly 77,000 were running software versions affected by at least one recently disclosed critical vulnerability. Researchers also detected over 54,000 instances of OpenClaw control panels left open to the web. This open-source agent allows users to link their personal accounts and various other services to the platform. The data does not confirm whether these specific deployments are actively vulnerable to exploitation. However, the findings highlight the risks associated with exposing software that has access to sensitive company systems. Austin advised that companies need to treat AI infrastructure like any other deployed software. This includes regularly reviewing, updating, and securing these systems to mitigate potential threats.