A lot of this will be in Unicode, so there may be no simple character set/encoding declaration to filter on. Unicode is a universal, one-size-fits-all encoding that supports many different human languages. I'd suggest you inspect the source (ctrl+u) of a few of these messages and hopefully you can prove me wrong; if they are indeed using specific encodings you could feasibly filter them, but you'll need to build up your own blacklist of unwelcome encodings. Look for "charset". I do have a filter that triggers on characters outside the usual "latin" range that English and European languages use. Say if this sounds of interest to you and I can post here what I use. Some users seem to think that we should all be using 7-bit US ASCII exclusively and they tend to view Unicode with suspicion, but the world is moving on and Unicode is becoming the de facto standard.