AbstractJsonpProvider currently uses Character.defaultCharset() if no charset is specified in the media type. E.g. on Windows, this causes the JSON parser to use a non-unicode charset for parsing the JSON. Since the default encoding for JSON is UTF-8 according to the specification, it would make more sense to strictly pass UTF-8 as the default charset to the parser or, alternatively, let the parser figure out the charset itself by constructing it entirely without charset.