Chinmay Gondhalekar

COMA: A Compositional Misleading Attack Class on Security-RAG, and a Causal Counterfactual Defense

An audit is proposed that measures the leave-one-out causal influence of each retrieved document and flags answers whose influence concentrates on low-trust documents, and proposes an audit that measures the leave-one-out causal influence of each retrieved document and flags answers whose influence concentrates on low-...