Timestamps are coerced to different types in 0.25.3 and 1.0.0 when selecting a row

[oalfred]

Code Sample, a copy-pastable example if possible

import pandas as pd
df = pd.DataFrame({'TIME': [pd.to_datetime('2020-01-01 12:00')], 'STRING': ['ABC']})
print(type(df.iloc[0]['TIME']))

Problem description

In 0.25.3 the code above results in <class 'pandas._libs.tslibs.timestamps.Timestamp'> whereas in 1.0.0 it is numpy.datetime64. I could not find anything about this specifically in the release notes. Is it intended?

[jreback]

looks the same as #31649

which is already patched for 1.0.1

cc @jbrockmendel

[TomAugspurger]

Yeah, thanks for the report @oalfred, but this has been fixed already.