After upgrading from Apex 5 to 23.2 users have reported odd characters in data they upload to other third party applications from CSV files they download from our APEX applications. The third party applications require the CSV files to be in UTF8 format. The new “Automatic CSV Encoding” settings appears to be setting the encoding to "Western (Windows 1252)". Turning this off makes no difference, the CSV files still use the same encoding. The NLS_CHARACTERSET of the underlying database is WE8MSWIN1252.
How can I force the downloaded CSV files to be encoded as UTF8?