diff --git a/kernels/portable/cpu/op_constant_pad_nd.cpp b/kernels/portable/cpu/op_constant_pad_nd.cpp
index 6e643e1b945..be3962e018c 100644
--- a/kernels/portable/cpu/op_constant_pad_nd.cpp
+++ b/kernels/portable/cpu/op_constant_pad_nd.cpp
@@ -185,7 +185,10 @@ Tensor& constant_pad_nd_out(
   ScalarType in_type = in.scalar_type();
 
   ET_SWITCH_REALHBBF16_TYPES(in_type, ctx, "constant_pad_nd.out", CTYPE, [&]() {
-    const CTYPE value_casted = utils::scalar_to<CTYPE>(value);
+    auto opt_value_casted =
+        utils::internal::check_overflow_scalar_cast<CTYPE>(value);
+    ET_KERNEL_CHECK(ctx, opt_value_casted.has_value(), InvalidArgument, );
+    auto value_casted = opt_value_casted.value();
     constant_pad_nd_out_impl<CTYPE>(in, pad, value_casted, out);
   });
 
diff --git a/kernels/portable/cpu/op_full.cpp b/kernels/portable/cpu/op_full.cpp
index 83ffcad45a6..b83637f2b91 100644
--- a/kernels/portable/cpu/op_full.cpp
+++ b/kernels/portable/cpu/op_full.cpp
@@ -37,7 +37,10 @@ Tensor& full_out(
   constexpr auto name = "full.out";
 
   ET_SWITCH_REALHBBF16_TYPES(out_type, ctx, name, CTYPE_OUT, [&] {
-    CTYPE_OUT val_casted = utils::scalar_to<CTYPE_OUT>(fill_value);
+    auto opt_val_casted =
+        utils::internal::check_overflow_scalar_cast<CTYPE_OUT>(fill_value);
+    ET_KERNEL_CHECK(ctx, opt_val_casted.has_value(), InvalidArgument, );
+    auto val_casted = opt_val_casted.value();
     auto data_out = out.mutable_data_ptr<CTYPE_OUT>();
     for (const auto i : c10::irange(out.numel())) {
       data_out[i] = val_casted;
diff --git a/kernels/portable/cpu/op_full_like.cpp b/kernels/portable/cpu/op_full_like.cpp
index 7671cd61ea9..0e263fb9c10 100644
--- a/kernels/portable/cpu/op_full_like.cpp
+++ b/kernels/portable/cpu/op_full_like.cpp
@@ -48,23 +48,19 @@ Tensor& full_like_out(
       out,
       "Failed to resize output tensor.");
 
-  ScalarType val_type = utils::get_scalar_dtype(fill_value);
   ScalarType out_type = out.scalar_type();
 
   constexpr auto name = "scalar_tensor.out";
 
-  ET_SWITCH_REALB_TYPES(val_type, ctx, name, CTYPE_VAL, [&] {
-    CTYPE_VAL val;
-    ET_KERNEL_CHECK(
-        ctx, utils::extract_scalar(fill_value, &val), InvalidArgument, );
-
-    ET_SWITCH_REALHBBF16_TYPES(out_type, ctx, name, CTYPE_OUT, [&] {
-      CTYPE_OUT val_casted = static_cast<CTYPE_OUT>(val);
-      auto data_out = out.mutable_data_ptr<CTYPE_OUT>();
-      for (const auto i : c10::irange(out.numel())) {
-        data_out[i] = val_casted;
-      }
-    });
+  ET_SWITCH_REALHBBF16_TYPES(out_type, ctx, name, CTYPE_OUT, [&] {
+    auto opt_val_casted =
+        utils::internal::check_overflow_scalar_cast<CTYPE_OUT>(fill_value);
+    ET_KERNEL_CHECK(ctx, opt_val_casted.has_value(), InvalidArgument, );
+    auto val_casted = opt_val_casted.value();
+    auto data_out = out.mutable_data_ptr<CTYPE_OUT>();
+    for (const auto i : c10::irange(out.numel())) {
+      data_out[i] = val_casted;
+    }
   });
 
   return out;
diff --git a/kernels/portable/cpu/op_hardtanh.cpp b/kernels/portable/cpu/op_hardtanh.cpp
index 8ec73b07856..65411d5f6b0 100644
--- a/kernels/portable/cpu/op_hardtanh.cpp
+++ b/kernels/portable/cpu/op_hardtanh.cpp
@@ -45,8 +45,15 @@ Tensor& hardtanh_out(
   ET_KERNEL_CHECK(ctx, in_type == out_type, InvalidArgument, out);
 
   ET_SWITCH_REALHBF16_TYPES(in_type, ctx, "hardtanh.out", CTYPE, [&]() {
-    const CTYPE min_casted = utils::scalar_to<CTYPE>(min);
-    const CTYPE max_casted = utils::scalar_to<CTYPE>(max);
+    auto opt_min_casted =
+        utils::internal::check_overflow_scalar_cast<CTYPE>(min);
+    ET_KERNEL_CHECK(ctx, opt_min_casted.has_value(), InvalidArgument, );
+    auto min_casted = opt_min_casted.value();
+
+    auto opt_max_casted =
+        utils::internal::check_overflow_scalar_cast<CTYPE>(max);
+    ET_KERNEL_CHECK(ctx, opt_max_casted.has_value(), InvalidArgument, );
+    auto max_casted = opt_max_casted.value();
 
     apply_unary_map_fn(
         [min_casted, max_casted](const CTYPE val_in) {
diff --git a/kernels/portable/cpu/op_leaky_relu.cpp b/kernels/portable/cpu/op_leaky_relu.cpp
index 11860c8d129..fa62a75974e 100644
--- a/kernels/portable/cpu/op_leaky_relu.cpp
+++ b/kernels/portable/cpu/op_leaky_relu.cpp
@@ -44,7 +44,11 @@ Tensor& leaky_relu_out(
   ET_KERNEL_CHECK(ctx, in_type == out_type, InvalidArgument, out);
 
   ET_SWITCH_FLOATHBF16_TYPES(in_type, ctx, "leaky_relu.out", CTYPE, [&]() {
-    const CTYPE negative_slope_casted = utils::scalar_to<CTYPE>(negative_slope);
+    auto opt_negative_slope_casted =
+        utils::internal::check_overflow_scalar_cast<CTYPE>(negative_slope);
+    ET_KERNEL_CHECK(
+        ctx, opt_negative_slope_casted.has_value(), InvalidArgument, );
+    auto negative_slope_casted = opt_negative_slope_casted.value();
 
     apply_unary_map_fn(
         [negative_slope_casted](const CTYPE val_in) {
diff --git a/kernels/portable/cpu/op_scalar_tensor.cpp b/kernels/portable/cpu/op_scalar_tensor.cpp
index e111a9ac869..bff4ecc318c 100644
--- a/kernels/portable/cpu/op_scalar_tensor.cpp
+++ b/kernels/portable/cpu/op_scalar_tensor.cpp
@@ -24,17 +24,11 @@ scalar_tensor_out(KernelRuntimeContext& ctx, const Scalar& s, Tensor& out) {
 
   constexpr auto name = "scalar_tensor.out";
 
-  if (s.isFloatingPoint() &&
-      executorch::runtime::isIntegralType(out_type, false)) {
-    ET_SWITCH_INT_TYPES(out_type, ctx, name, CTYPE, [&]() {
-      out.mutable_data_ptr<CTYPE>()[0] =
-          static_cast<CTYPE>(utils::scalar_to<int64_t>(s));
-    });
-  } else {
-    ET_SWITCH_REALHBBF16_TYPES(out_type, ctx, name, CTYPE, [&]() {
-      out.mutable_data_ptr<CTYPE>()[0] = utils::scalar_to<CTYPE>(s);
-    });
-  }
+  ET_SWITCH_REALHBBF16_TYPES(out_type, ctx, name, CTYPE, [&]() {
+    auto opt_val_casted = utils::internal::check_overflow_scalar_cast<CTYPE>(s);
+    ET_KERNEL_CHECK(ctx, opt_val_casted.has_value(), InvalidArgument, );
+    out.mutable_data_ptr<CTYPE>()[0] = opt_val_casted.value();
+  });
 
   return out;
 }
diff --git a/kernels/portable/cpu/op_scatter.cpp b/kernels/portable/cpu/op_scatter.cpp
index 7de0ec4d5f9..965afbb4b66 100644
--- a/kernels/portable/cpu/op_scatter.cpp
+++ b/kernels/portable/cpu/op_scatter.cpp
@@ -154,7 +154,9 @@ Tensor& scatter_value_out(
   constexpr auto name = "scatter.value_out";
 
   ET_SWITCH_REALHBBF16_TYPES(in.scalar_type(), ctx, name, CTYPE, [&]() {
-    const CTYPE val = utils::scalar_to<CTYPE>(value);
+    auto opt_val = utils::internal::check_overflow_scalar_cast<CTYPE>(value);
+    ET_KERNEL_CHECK(ctx, opt_val.has_value(), InvalidArgument, );
+    auto val = opt_val.value();
     scatter_value_helper<CTYPE>(in, dim, index, val, out);
   });
 
diff --git a/kernels/test/op_constant_pad_nd_test.cpp b/kernels/test/op_constant_pad_nd_test.cpp
index 88bee1d0ad9..7f44068d9cb 100644
--- a/kernels/test/op_constant_pad_nd_test.cpp
+++ b/kernels/test/op_constant_pad_nd_test.cpp
@@ -7,6 +7,7 @@
  */
 
 #include <executorch/kernels/test/FunctionHeaderWrapper.h> // Declares the operator
+#include <executorch/kernels/test/ScalarOverflowTestMacros.h>
 #include <executorch/kernels/test/TestUtil.h>
 #include <executorch/kernels/test/supported_features.h>
 #include <executorch/runtime/core/exec_aten/exec_aten.h>
@@ -347,6 +348,21 @@ class OpConstantPadNDOutTest : public OperatorTest {
     op_constant_pad_nd_out(self, padding_ref, 7, out);
     EXPECT_TENSOR_CLOSE(out, expected);
   }
+
+  template <ScalarType DTYPE>
+  void expect_bad_scalar_value_dies(const Scalar& bad_value) {
+    TensorFactory<DTYPE> tf;
+    const std::vector<int32_t> sizes = {2, 2};
+    const std::vector<int32_t> sizes_out = {2, 4};
+    const std::vector<int64_t> padding = {1, 1};
+
+    IntArrayRef padding_ref = IntArrayRef(padding.data(), padding.size());
+    Tensor self = tf.ones(sizes);
+    Tensor out = tf.zeros(sizes_out);
+
+    ET_EXPECT_KERNEL_FAILURE(
+        context_, op_constant_pad_nd_out(self, padding_ref, bad_value, out));
+  }
 };
 
 TEST_F(OpConstantPadNDOutTest, TestPadDim2) {
@@ -465,3 +481,5 @@ TEST_F(OpConstantPadNDOutTest, IncorrectOutputShapeFail) {
   ET_EXPECT_KERNEL_FAILURE(
       context_, op_constant_pad_nd_out(self, padding_ref, 0, out));
 }
+
+GENERATE_SCALAR_OVERFLOW_TESTS(OpConstantPadNDOutTest)
diff --git a/kernels/test/op_full_like_test.cpp b/kernels/test/op_full_like_test.cpp
index c0b9dcc4107..6e7692f5347 100644
--- a/kernels/test/op_full_like_test.cpp
+++ b/kernels/test/op_full_like_test.cpp
@@ -7,6 +7,7 @@
  */
 
 #include <executorch/kernels/test/FunctionHeaderWrapper.h> // Declares the operator
+#include <executorch/kernels/test/ScalarOverflowTestMacros.h>
 #include <executorch/kernels/test/TestUtil.h>
 #include <executorch/kernels/test/supported_features.h>
 #include <executorch/runtime/core/exec_aten/exec_aten.h>
@@ -65,6 +66,18 @@ class OpFullLikeTest : public OperatorTest {
     ET_EXPECT_KERNEL_FAILURE(
         context_, op_full_like_out(in, value, memory_format, out));
   }
+
+  template <ScalarType DTYPE>
+  void expect_bad_scalar_value_dies(const Scalar& bad_value) {
+    TensorFactory<DTYPE> tf;
+    const std::vector<int32_t> sizes = {2, 2};
+    Tensor in = tf.zeros(sizes);
+    Tensor out = tf.zeros(sizes);
+    optional<MemoryFormat> memory_format;
+
+    ET_EXPECT_KERNEL_FAILURE(
+        context_, op_full_like_out(in, bad_value, memory_format, out));
+  }
 };
 
 template <>
@@ -209,3 +222,5 @@ TEST_F(OpFullLikeTest, HalfSupport) {
   op_full_like_out(in, INFINITY, memory_format, out);
   EXPECT_TENSOR_CLOSE(out, tf.full({2, 3}, INFINITY));
 }
+
+GENERATE_SCALAR_OVERFLOW_TESTS(OpFullLikeTest)
diff --git a/kernels/test/op_full_test.cpp b/kernels/test/op_full_test.cpp
index 93129679087..35115dc7ed6 100644
--- a/kernels/test/op_full_test.cpp
+++ b/kernels/test/op_full_test.cpp
@@ -7,6 +7,7 @@
  */
 
 #include <executorch/kernels/test/FunctionHeaderWrapper.h> // Declares the operator
+#include <executorch/kernels/test/ScalarOverflowTestMacros.h>
 #include <executorch/kernels/test/TestUtil.h>
 #include <executorch/kernels/test/supported_features.h>
 #include <executorch/runtime/core/exec_aten/exec_aten.h>
@@ -59,6 +60,17 @@ class OpFullOutTest : public OperatorTest {
     op_full_out(aref, 1.0, out);
     EXPECT_TENSOR_EQ(out, tf.ones(size_int32_t));
   }
+
+  template <ScalarType DTYPE>
+  void expect_bad_scalar_value_dies(const Scalar& bad_value) {
+    TensorFactory<DTYPE> tf;
+    std::vector<int32_t> sizes = {2, 2};
+    std::vector<int64_t> sizes_int64_t(sizes.begin(), sizes.end());
+    auto aref = IntArrayRef(sizes_int64_t.data(), sizes_int64_t.size());
+    Tensor out = tf.zeros(sizes);
+
+    ET_EXPECT_KERNEL_FAILURE(context_, op_full_out(aref, bad_value, out));
+  }
 };
 
 #define GENERATE_TEST(_, DTYPE)                  \
@@ -72,20 +84,7 @@ class OpFullOutTest : public OperatorTest {
 
 ET_FORALL_REALHBF16_TYPES(GENERATE_TEST)
 
-TEST_F(OpFullOutTest, ValueOverflow) {
-  if (torch::executor::testing::SupportedFeatures::get()->is_aten) {
-    GTEST_SKIP() << "ATen kernel doesn't handle overflow";
-  }
-  TensorFactory<ScalarType::Byte> tf;
-
-  std::vector<int64_t> sizes_int64_t_vec = {2, 3};
-  std::vector<int32_t> sizes_in32_t_vec = {2, 3};
-  auto sizes = IntArrayRef(sizes_int64_t_vec.data(), sizes_int64_t_vec.size());
-
-  Tensor out = tf.zeros(sizes_in32_t_vec);
-
-  op_full_out(sizes, 1000, out);
-}
+GENERATE_SCALAR_OVERFLOW_TESTS(OpFullOutTest)
 
 TEST_F(OpFullOutTest, HalfSupport) {
   TensorFactory<ScalarType::Half> tf;
diff --git a/kernels/test/op_hardtanh_test.cpp b/kernels/test/op_hardtanh_test.cpp
index 72d09063d3e..38b0eeea40f 100644
--- a/kernels/test/op_hardtanh_test.cpp
+++ b/kernels/test/op_hardtanh_test.cpp
@@ -7,6 +7,7 @@
  */
 
 #include <executorch/kernels/test/FunctionHeaderWrapper.h> // Declares the operator
+#include <executorch/kernels/test/ScalarOverflowTestMacros.h>
 #include <executorch/kernels/test/TestUtil.h>
 #include <executorch/runtime/core/exec_aten/exec_aten.h>
 #include <executorch/runtime/core/exec_aten/testing_util/tensor_factory.h>
@@ -51,6 +52,21 @@ class OpHardTanhTest : public OperatorTest {
     EXPECT_TENSOR_EQ(out, ret);
     EXPECT_TENSOR_EQ(out, tf.make({2, 2}, {lower_bound, 0, 1, 2}));
   }
+
+  template <ScalarType DTYPE>
+  void expect_bad_scalar_value_dies(const Scalar& bad_value) {
+    TensorFactory<DTYPE> tf;
+    Tensor in = tf.ones({2, 2});
+    Tensor out = tf.zeros({2, 2});
+
+    // Test overflow for min parameter (using valid max)
+    ET_EXPECT_KERNEL_FAILURE(
+        context_, op_hardtanh_out(in, bad_value, 1.0, out));
+
+    // Test overflow for max parameter (using valid min)
+    ET_EXPECT_KERNEL_FAILURE(
+        context_, op_hardtanh_out(in, -1.0, bad_value, out));
+  }
 };
 
 TEST_F(OpHardTanhTest, SanityCheck) {
@@ -58,3 +74,5 @@ TEST_F(OpHardTanhTest, SanityCheck) {
   ET_FORALL_REALHBF16_TYPES(TEST_ENTRY);
 #undef TEST_ENTRY
 }
+
+GENERATE_SCALAR_OVERFLOW_TESTS(OpHardTanhTest)
diff --git a/kernels/test/op_leaky_relu_test.cpp b/kernels/test/op_leaky_relu_test.cpp
index 847c00652be..6b2e3083e2e 100644
--- a/kernels/test/op_leaky_relu_test.cpp
+++ b/kernels/test/op_leaky_relu_test.cpp
@@ -40,6 +40,15 @@ class OpLeakyReluTest : public OperatorTest {
     EXPECT_TENSOR_EQ(out, ret);
     EXPECT_TENSOR_EQ(out, tf.ones({2, 2}));
   }
+
+  template <ScalarType DTYPE>
+  void expect_bad_scalar_value_dies(const Scalar& bad_value) {
+    TensorFactory<DTYPE> tf;
+    Tensor in = tf.ones({2, 2});
+    Tensor out = tf.zeros({2, 2});
+
+    ET_EXPECT_KERNEL_FAILURE(context_, op_leaky_relu_out(in, bad_value, out));
+  }
 };
 
 TEST_F(OpLeakyReluTest, SanityCheck) {
@@ -47,3 +56,13 @@ TEST_F(OpLeakyReluTest, SanityCheck) {
   ET_FORALL_FLOATHBF16_TYPES(TEST_ENTRY);
 #undef TEST_ENTRY
 }
+
+TEST_F(OpLeakyReluTest, FloatTensorTooSmallScalarDies) {
+  /* Cannot be represented by a float. */
+  expect_bad_scalar_value_dies<ScalarType::Float>(-3.41e+38);
+}
+
+TEST_F(OpLeakyReluTest, FloatTensorTooLargeScalarDies) {
+  /* Cannot be represented by a float. */
+  expect_bad_scalar_value_dies<ScalarType::Float>(3.41e+38);
+}
diff --git a/kernels/test/op_scalar_tensor_test.cpp b/kernels/test/op_scalar_tensor_test.cpp
index db4816e8847..0be6f395eb0 100644
--- a/kernels/test/op_scalar_tensor_test.cpp
+++ b/kernels/test/op_scalar_tensor_test.cpp
@@ -7,6 +7,7 @@
  */
 
 #include <executorch/kernels/test/FunctionHeaderWrapper.h> // Declares the operator
+#include <executorch/kernels/test/ScalarOverflowTestMacros.h>
 #include <executorch/kernels/test/TestUtil.h>
 #include <executorch/kernels/test/supported_features.h>
 #include <executorch/runtime/core/exec_aten/exec_aten.h>
@@ -71,6 +72,14 @@ class OpScalarTensorOutTest : public OperatorTest {
 
     ET_EXPECT_KERNEL_FAILURE(context_, op_scalar_tensor_out(value, out));
   }
+
+  template <ScalarType DTYPE>
+  void expect_bad_scalar_value_dies(const Scalar& bad_value) {
+    TensorFactory<DTYPE> tf;
+    Tensor out = tf.zeros({});
+
+    ET_EXPECT_KERNEL_FAILURE(context_, op_scalar_tensor_out(bad_value, out));
+  }
 };
 
 #define GENERATE_TEST_0D(ctype, dtype)                      \
@@ -131,3 +140,5 @@ TEST_F(OpScalarTensorOutTest, HalfSupport) {
   op_scalar_tensor_out(INFINITY, out);
   EXPECT_TENSOR_CLOSE(out, tf.make({}, {INFINITY}));
 }
+
+GENERATE_SCALAR_OVERFLOW_TESTS(OpScalarTensorOutTest)
diff --git a/kernels/test/op_scatter_test.cpp b/kernels/test/op_scatter_test.cpp
index 0e55aadaeda..dac9017d188 100644
--- a/kernels/test/op_scatter_test.cpp
+++ b/kernels/test/op_scatter_test.cpp
@@ -7,6 +7,7 @@
  */
 
 #include <executorch/kernels/test/FunctionHeaderWrapper.h> // Declares the operator
+#include <executorch/kernels/test/ScalarOverflowTestMacros.h>
 #include <executorch/kernels/test/TestUtil.h>
 #include <executorch/kernels/test/supported_features.h>
 #include <executorch/runtime/core/exec_aten/exec_aten.h>
@@ -364,6 +365,19 @@ class OpScatterValueOutTest : public OperatorTest {
     op_scatter_value_out(input, 2, index, value, out);
     EXPECT_TENSOR_EQ(out, expected);
   }
+
+  template <ScalarType DTYPE>
+  void expect_bad_scalar_value_dies(const Scalar& bad_value) {
+    TensorFactory<DTYPE> tf;
+    TensorFactory<ScalarType::Long> tf_index;
+
+    Tensor self = tf.ones({2, 2});
+    Tensor index = tf_index.zeros({2, 2});
+    Tensor out = tf.zeros({2, 2});
+
+    ET_EXPECT_KERNEL_FAILURE(
+        context_, op_scatter_value_out(self, 0, index, bad_value, out));
+  }
 };
 
 TEST_F(OpScatterSrcOutTest, AllValidInputOutputSupport) {
@@ -652,3 +666,5 @@ TEST_F(OpScatterSrcOutTest, InvalidOneDimInputAndZeroDimIndex) {
   ET_EXPECT_KERNEL_FAILURE(
       context_, op_scatter_src_out(self, 0, index, src, out));
 }
+
+GENERATE_SCALAR_OVERFLOW_TESTS(OpScatterValueOutTest)