Why can't the comm command tell the difference between ✓&⨯ when LANG=en_US.UTF-8?

If I have two files, a.txt containing:

meow ✓ bar

and b.txt containing:

meow ⨯ bar

Why can't the GNU Coreutils command comm tell the difference between the two when $LANG is set to a UTF-8 encoding?

$ LANG=C comm a.txt b.txtmeow ✓ bar    meow ⨯ bar$ LANG=en_US comm a.txt b.txtmeow ✓ bar    meow ⨯ bar$ LANG=en_US.UTF-8 comm a.txt b.txt        meow ⨯ bar