A Small Language Model is an AI model built for a narrower set of tasks than a large language model, usually with fewer parameters and lower compute, memory, and energy requirements. SLMs are often tuned for a specific domain, such as customer support, document classification, or security triage, where focused behavior matters more than broad general knowledge.
In cyber security, SLMs matter because they can be deployed closer to sensitive data, including on private servers, edge systems, or even inside trusted hardware environments. That can reduce exposure compared with sending prompts to external services and can make monitoring, audit, and access control easier. Defenders use SLMs for tasks like phishing detection, log analysis, and policy lookup. Attackers may also use them to automate spam, generate lure text, or scale reconnaissance cheaply. Their narrower scope is useful, but it also means they need careful validation, prompt controls, and ongoing monitoring for drift and misuse.