diff --git a/CHANGELOG.md b/CHANGELOG.md
index 97fd31e5a50..d109d977d96 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -89,6 +89,7 @@ This release is compatible with NumPy 2.5.
 * Fixed `dpnp.ndarray.view` ignoring the USM element offset of a sliced array, which also caused `dpnp.einsum` to silently return wrong results for a single sliced operand with no summed index [#3037](https://github.com/IntelPython/dpnp/pull/3037)
 * Fixed `dpnp.all` and `dpnp.any` aborting when reducing over an empty axis (e.g. an array with a zero-length dimension) [#3021](https://github.com/IntelPython/dpnp/pull/3021)
 * Released the GIL before the blocking OneMKL DFT calls in the FFT extension [#3040](https://github.com/IntelPython/dpnp/pull/3040)
+* Fixed `dpnp.insert` silently ignoring out-of-bounds negative indices in a multi-element `obj`, so a mix of in-bounds and out-of-bounds indices now consistently raises `IndexError` [#3041](https://github.com/IntelPython/dpnp/pull/3041)
 
 ### Security
 
diff --git a/dpnp/dpnp_iface_manipulation.py b/dpnp/dpnp_iface_manipulation.py
index b2046ffc494..53cbd231a89 100644
--- a/dpnp/dpnp_iface_manipulation.py
+++ b/dpnp/dpnp_iface_manipulation.py
@@ -254,6 +254,15 @@ def _insert_array_indices(parameters, indices, values, obj):
         # Can safely cast the empty list to intp
         indices = indices.astype(dpnp.intp)
 
+    if indices.size > 0:
+        min_idx = int(indices.min())
+        max_idx = int(indices.max())
+        if min_idx < -n or max_idx > n:
+            oob = min_idx if min_idx < -n else max_idx
+            raise IndexError(
+                f"index {oob} is out of bounds for axis {axis} with size {n}"
+            )
+
     indices[indices < 0] += n
 
     numnew = len(indices)
diff --git a/dpnp/tests/test_manipulation.py b/dpnp/tests/test_manipulation.py
index 3dbd9691d4c..7245c6712a3 100644
--- a/dpnp/tests/test_manipulation.py
+++ b/dpnp/tests/test_manipulation.py
@@ -815,11 +815,32 @@ def test_error(self):
         with pytest.raises(TypeError):
             dpnp.insert(a, [], 2, axis="nonsense")
 
-    @pytest.mark.parametrize("idx", [4, -4])
-    def test_index_out_of_bounds(self, idx):
-        a = dpnp.array([0, 1, 2])
+    @testing.with_requires("numpy>=2.6")
+    @pytest.mark.parametrize("xp", [numpy, dpnp])
+    @pytest.mark.parametrize(
+        "idx, values",
+        [
+            # single-element obj -> singleton path
+            ([4], [3, 4]),
+            ([-4], [3, 4]),
+            # multi-element obj -> array path
+            ([-6, 0], [9, 8]),
+            ([0, 6], [9, 8]),
+            ([4, 4], [3, 4]),
+            ([-4, -5], [3, 4]),
+        ],
+    )
+    def test_index_out_of_bounds(self, xp, idx, values):
+        a = xp.array([0, 1, 2])
+        with pytest.raises(IndexError, match="out of bounds"):
+            xp.insert(a, idx, values)
+
+    @pytest.mark.parametrize("xp", [numpy, dpnp])
+    @pytest.mark.parametrize("axis", [0, 1])
+    def test_index_out_of_bounds_ndim(self, xp, axis):
+        a = xp.ones((3, 3))
         with pytest.raises(IndexError, match="out of bounds"):
-            dpnp.insert(a, [idx], [3, 4])
+            xp.insert(a, [5, 0], 9, axis=axis)
 
 
 # array_split has more comprehensive test of splitting.