Security Becomes the Control Plane for Enterprise AI Factories
Enterprise AI factories are creating a new attack surface that traditional cybersecurity was not built to handle. Dell and Intel are embedding security into infrastructure as agent identity, model integrity and attack speed become critical, SiliconANGLE Media reports.
AI factories turn companies into factories of intelligence, but the model they use was not designed with security at its center, according to SiliconANGLE Media. Energy, compute and data go in, and intelligence in the form of tokens comes out. New risks come out as well, and traditional cybersecurity was never built to handle them.
Dave Vellante, chief analyst at theCUBE Research, said in a recent analysis that AI factories introduce a new class of risks that extend beyond traditional cybersecurity models. In this world, data is not static, bounded or easily classified, he said. Instead, it is continuously generated, transformed and consumed across distributed environments.
Vellante explained that models are supported by deterministic systems that behave predictably, but they are also adaptive, probabilistic engines that evolve over time. Agents increasingly act autonomously, interacting with enterprise systems, executing workflows and making decisions with limited or sometimes no human intervention, he said.
Dell and Intel are responding by building security into the infrastructure itself and trying to establish data protection from the ground up, SiliconANGLE Media reported. Steve Kenniston, senior cybersecurity evangelist for portfolio marketing at Dell, said AI changes the whole game. He pointed to model inferencing, training models and training data, systems vulnerable to prompt injection, and identity management that must be considered. Every new application has a new attack surface, Kenniston said.
Mukund Khatri, fellow and vice president of systems architecture at Dell, said AI brings additional threats. The model and large language models look like code, he said, and they need to be protected like code, but they are essentially data. Model integrity is of paramount importance, Khatri added.
AI agents are becoming more common across enterprise environments, creating new security questions as they gain the ability to act autonomously. Because agents are created and managed by humans, organizations must consider how their actions and access could put sensitive data and systems at risk. Khatri said large language models could give wrong answers, but agents could take the wrong action, which is more detrimental. When systems work autonomously, the risk of errors or malicious activity can increase, especially because actions are executed in real time, often without an opportunity for human review.
Identity has become a critical control point for AI agents. Businesses are adopting AI at high speed, but it is not clear how agents operate and what they can access. Agent identity has become a basic layer of control because poorly governed agents could provide an opening for bad actors. According to the Darktrace 2026 State of AI Cybersecurity report, which surveyed 1,540 cybersecurity leaders and practitioners across 14 countries, 92 percent said they were concerned about the security implications of AI agents across their workforce, while 46 percent said they did not feel adequately prepared to defend against AI-powered threats. The report also found that 87 percent believe AI is significantly increasing the sophistication and success rate of malware.
Traditional security models built around perimeter defense, static policy enforcement and human-speed response are increasingly insufficient, according to SiliconANGLE Media. Practices such as model inversion, prompt injection and data poisoning can exploit weaknesses involving models and data rather than conventional software vulnerabilities alone. Knowing where the danger lies is only half the problem; the other half is speed.
AI can help adversaries identify vulnerabilities and develop attacks more quickly, shrinking the time between vulnerability discovery and active exploitation in ways traditional enterprise security frameworks were not designed to address. AI models can analyze software, identify potential flaws and accelerate parts of the attack process. When agents move through systems at high speed, security becomes increasingly difficult to maintain, and AI workloads bring a complexity that most security teams are not ready to handle. To make the use of agents safer, least-privilege access will be crucial, and governance concepts such as lifecycle management are part of the discussion, according to the article.