MIT Lincoln Laboratory focuses on researching and developing technologies to support national security. Given the ubiquity of AI, the team explored making AI systems trustworthy and reliable through topological data analysis (TDA). TDA uses topological and geometric properties to characterize both global and local structures within high-dimensional data. Therefore, two problems were attempted: (1) interpreting topological features extracted from text linguistically; and (2) observing the topological effects of adversarial perturbations in language models using attention maps.