BUG: setting dt64 values into Series[int] incorrectly casting dt64->int

doc/source/whatsnew/v1.3.0.rst

@@ -281,6 +281,7 @@ Indexing
 - Bug in :meth:`DataFrame.loc` dropping levels of :class:`MultiIndex` when :class:`DataFrame` used as input has only one row (:issue:`10521`)
 - Bug in setting ``timedelta64`` values into numeric :class:`Series` failing to cast to object dtype (:issue:`39086`)
 - Bug in setting :class:`Interval` values into a :class:`Series` or :class:`DataFrame` with mismatched :class:`IntervalDtype` incorrectly casting the new values to the existing dtype (:issue:`39120`)
+- Bug in setting ``datetime64`` values into a :class:`Series` with integer-dtype incorrect casting the datetime64 values to integers (:issue:`39266`)
 
 Missing
 ^^^^^^^

pandas/core/internals/blocks.py

@@ -991,6 +991,18 @@ def setitem(self, indexer, value):
 
         # set
         else:
+            if (
+                self.is_object
+                and not is_ea_value
+                and arr_value.dtype.kind in ["m", "M"]
+            ):
+                # https://github.com/numpy/numpy/issues/12550
+                #  numpy will incorrect cast to int if we're not careful
+                if is_list_like(value):
+                    value = list(value)
+                else:
+                    value = [value] * len(values[indexer])
+
             values[indexer] = value
 
         if transpose:

pandas/tests/series/indexing/test_setitem.py

@@ -239,6 +239,27 @@ def test_setitem_callable_other(self):
         tm.assert_series_equal(ser, expected)
 
 
+class TestSetitemCasting:
+    @pytest.mark.parametrize("dtype", ["M8[ns]", "m8[ns]"])
+    def test_setitem_dt64_into_int_series(self, dtype):
+        # dont cast dt64 to int when doing this setitem
+        orig = Series([1, 2, 3])
+
+        val = np.datetime64("2021-01-18 13:25:00", "ns")
+        if dtype == "m8[ns]":
+            val = val - val
+
+        ser = orig.copy()
+        ser[:-1] = val
+        expected = Series([val, val, 3], dtype=object)
+        tm.assert_series_equal(ser, expected)
+        assert isinstance(ser[0], type(val))
+
+        ser = orig.copy()
+        ser[:-1] = np.array([val, val])
+        tm.assert_series_equal(ser, expected)
+
+
 @pytest.mark.parametrize(
     "obj,expected,key",
     [