diff --git a/src/promptflow-evals/promptflow/evals/_constants.py b/src/promptflow-evals/promptflow/evals/_constants.py
index 38d19028c0f..893f2f06bbd 100644
--- a/src/promptflow-evals/promptflow/evals/_constants.py
+++ b/src/promptflow-evals/promptflow/evals/_constants.py
@@ -19,7 +19,7 @@ class EvaluationMetrics:
     SELF_HARM = "self_harm"
     SEXUAL = "sexual"
     PROTECTED_MATERIAL = "protected_material"
-    INDIRECT_ATTACK = "indirect_attack"
+    XPIA = "xpia"
 
 
 class _InternalEvaluationMetrics:
diff --git a/src/promptflow-evals/promptflow/evals/evaluate/_evaluate.py b/src/promptflow-evals/promptflow/evals/evaluate/_evaluate.py
index 4f015c912b4..578236c527f 100644
--- a/src/promptflow-evals/promptflow/evals/evaluate/_evaluate.py
+++ b/src/promptflow-evals/promptflow/evals/evaluate/_evaluate.py
@@ -92,7 +92,7 @@ def _aggregate_label_defect_metrics(df: pd.DataFrame) -> Tuple[List[str], Dict[s
     handled_metrics = [
         EvaluationMetrics.PROTECTED_MATERIAL,
         _InternalEvaluationMetrics.ECI,
-        EvaluationMetrics.INDIRECT_ATTACK,
+        EvaluationMetrics.XPIA,
     ]
     label_cols = []
     for col in df.columns: