2026. Path-Aware Defense Against Prompt Injection Attacks in Retrieval-Augmented Large Language Models. Journal of Advanced Artificial Intelligence Research. 1, 1 (Jun. 2026).