Interpretable Safety Routing for Foundation Models in Commercial Recommendation Pipelines Using Path-Level Causal Intervention. (2026). Journal of Advanced Artificial Intelligence Research, 5(1). https://www.jaair.org/index.php/home/article/view/106