The Pentagon is planning for AI companies to train on classified data, defense (technologyreview.com)

And everyone knows that the Pentagon would never make a mistake

If these companies are training on classified defense data, how do they guarantee the model won't inadvertently leak tactical specifics through a clever jailbreak or even just a nuanced inference? The boundary between 'learning a pattern' and 'storing a secret' is still too blurry for comfort in a defense context.