[1]
2026. Interpretable Safety Routing for Foundation Models in Commercial Recommendation Pipelines Using Path-Level Causal Intervention. Journal of Advanced Artificial Intelligence Research. 5, 1 (May 2026).