Small language models are more reliable and secure than their large counterparts, primarily because they draw information ...