diff --git a/onnxruntime/python/tools/transformers/io_binding_helper.py b/onnxruntime/python/tools/transformers/io_binding_helper.py
index 31e04216a8..3182107cd8 100644
--- a/onnxruntime/python/tools/transformers/io_binding_helper.py
+++ b/onnxruntime/python/tools/transformers/io_binding_helper.py
@@ -32,7 +32,7 @@ class TypeHelper:
             "tensor(int32)": numpy.intc,
             "tensor(float)": numpy.float32,
             "tensor(float16)": numpy.float16,
-            "tensor(bool)": numpy.bool,
+            "tensor(bool)": bool,
         }
         if ort_type not in ort_type_to_numpy_type_map:
             raise ValueError(f"{ort_type} not found in map")
@@ -61,7 +61,7 @@ class TypeHelper:
             numpy.int32: torch.int32,
             numpy.float32: torch.float32,
             numpy.float16: torch.float16,
-            numpy.bool: torch.bool,
+            bool: torch.bool,
         }
         if numpy_type not in numpy_type_to_torch_type_map:
             raise ValueError(f"{numpy_type} not found in map")
@@ -75,7 +75,7 @@ class TypeHelper:
             torch.int32: numpy.intc,
             torch.float32: numpy.float32,
             torch.float16: numpy.float16,
-            torch.bool: numpy.bool,
+            torch.bool: bool,
         }
         if torch_type not in torch_type_to_numpy_type_map:
             raise ValueError(f"{torch_type} not found in map")
diff --git a/onnxruntime/test/python/contrib_ops/onnx_test_torch_embedding.py b/onnxruntime/test/python/contrib_ops/onnx_test_torch_embedding.py
index f152962710..434331931d 100644
--- a/onnxruntime/test/python/contrib_ops/onnx_test_torch_embedding.py
+++ b/onnxruntime/test/python/contrib_ops/onnx_test_torch_embedding.py
@@ -84,7 +84,7 @@ class ONNXReferenceImplementationTest(unittest.TestCase):
         x = np.random.randn(3, 4).astype(np.int64)
         w = np.random.randn(10, 3).astype(np.float32)
         padding_idx = np.random.randint(3, size=1).astype(np.int64)
-        scale = np.array([1]).astype(np.bool)
+        scale = np.array([1]).astype(bool)
         y = torch_embedding_reference_implementation(w, x, padding_idx, scale)
         expect(
             node,
diff --git a/onnxruntime/test/testdata/transform/model_parallel/bart_mlp_megatron_basic_test.py b/onnxruntime/test/testdata/transform/model_parallel/bart_mlp_megatron_basic_test.py
index 8f8798750f..323ebf08e4 100644
--- a/onnxruntime/test/testdata/transform/model_parallel/bart_mlp_megatron_basic_test.py
+++ b/onnxruntime/test/testdata/transform/model_parallel/bart_mlp_megatron_basic_test.py
@@ -21,7 +21,7 @@ a_bias_initializer = numpy_helper.from_array(a_bias_np_vals, "encoder.t5_stack.b
 dropout_np_vals = np.asarray([0.1], dtype=np.float32).reshape(())
 dropout_initializer = numpy_helper.from_array(dropout_np_vals, "ratio")
 
-dropout_mode_np_vals = np.array([False], dtype=np.bool).reshape(())
+dropout_mode_np_vals = np.array([False], dtype=bool).reshape(())
 dropout_mode_initializer = numpy_helper.from_array(dropout_mode_np_vals, "mode")
 
 b_weight_np_vals = (0.01 * np.arange(hidden_size * weight_dim_to_split, dtype=np.float32)).reshape(
diff --git a/onnxruntime/test/testdata/transform/model_parallel/bart_self_attention_megatron_basic_test.py b/onnxruntime/test/testdata/transform/model_parallel/bart_self_attention_megatron_basic_test.py
index 8704079d1c..596b294ca2 100644
--- a/onnxruntime/test/testdata/transform/model_parallel/bart_self_attention_megatron_basic_test.py
+++ b/onnxruntime/test/testdata/transform/model_parallel/bart_self_attention_megatron_basic_test.py
@@ -66,7 +66,7 @@ where_shape_initializer = numpy_helper.from_array(
 dropout_np_vals = np.asarray([0.1], dtype=np.float32).reshape(())
 dropout_initializer = numpy_helper.from_array(dropout_np_vals, "ratio")
 
-dropout_mode_np_vals = np.array([False], dtype=np.bool).reshape(())
+dropout_mode_np_vals = np.array([False], dtype=bool).reshape(())
 dropout_mode_initializer = numpy_helper.from_array(dropout_mode_np_vals, "mode")
 
 shape_initializer3 = numpy_helper.from_array(
diff --git a/orttraining/orttraining/python/ort_trainer.py b/orttraining/orttraining/python/ort_trainer.py
index a9229bd3c3..a22bb78cd1 100644
--- a/orttraining/orttraining/python/ort_trainer.py
+++ b/orttraining/orttraining/python/ort_trainer.py
@@ -233,7 +233,7 @@ def dtype_torch_to_numpy(torch_dtype):
     elif torch_dtype == torch.int16 or torch_dtype == torch.short:
         return np.int16
     elif torch_dtype == torch.bool:
-        return np.bool
+        return bool
     else:
         raise Exception("Torch type to numpy type mapping unavailable for: " + str(torch_dtype))
 
diff --git a/orttraining/tools/scripts/opset12_model_transform.py b/orttraining/tools/scripts/opset12_model_transform.py
index b3cd82e470..c19aceb621 100644
--- a/orttraining/tools/scripts/opset12_model_transform.py
+++ b/orttraining/tools/scripts/opset12_model_transform.py
@@ -87,7 +87,7 @@ def process_trainabledropout(model):
             )
             index += 1
             # add training_mode output
-            mode_scalar = np.asarray([True]).astype(np.bool).reshape(())
+            mode_scalar = np.asarray([True]).astype(bool).reshape(())
             mode_value = numpy_helper.from_array(mode_scalar, "training_mode")
             training_mode_node = add_const(
                 model, "dropout_training_mode_node_%d" % index, "dropout_training_mode_%d" % index, t_value=mode_value