Language models like GPT are useful for analyzing the execution of commands. Most command structures in military communication would benefit from analysis to determine if someone is straying from command or moving toward violation of principles. The following published paper is a good example: Ethics, Rules of Engagement, and AI: Neural Narrative Mapping Using Large Transformer Language Models.

I think use of language models will be very beneficial to DOD agencies like NSA, but there is nothing to indicate what they have at this point.