I've noticed that the following file gets detected in a quite different way in chardet 3.0.3 compared to 2.3:

scriptencoding utf-8
" :purple_heart: 💜
" set list listchars=tab:»·,trail:·,eol:¬,nbsp:_,extends:❯,precedes:❮

chardet 2.3.0:
{'encoding': 'ISO-8859-2', 'confidence': 0.6680924803464797}

chardet 3.0.3:
{'encoding': 'Windows-1254', 'confidence': 0.5658124254347925, 'language': 'Turkish'}

This seems to be related to ISO-8859-2 being disabled currently, but I think it should be utf-8 after all?!
Remove the unicode glyph (purple heart, <💜> 128156, Hex 0001f49c, Octal 372234) makes it {'encoding': 'utf-8', 'confidence': 0.99, 'language': ''}.

(I've noticed this for the chardet.detect used in vint)