ESET unveils ESET NeoCortex cybersecurity AI model

ESET positions its NeoCortex model as purpose-built for security operations, contrasting it with general-purpose LLMs (large language models, AI that generates text) that are designed to produce plausible answers across broad topics. The company says security work instead demands consistent analysis of incident targets, timing, threat indicators, process relationships, and incomplete observations — the gap LiveCortex is meant to fill. Alongside that, ESET stresses that AI itself is an attack target: NeoCortex is designed to operate under strict safety measures, including a restricted sandbox, read-only tools, continuous monitoring, clear usage and token caps, a kill switch, and expert oversight for critical responses. ESET also says the model is developed and operated on its own infrastructure, which it says reduces reliance on external AI providers while keeping sensitive security data and model operations under its control.