Warn about potential homograph attacks
Homograph attacks exploit visual similarities of Unicode characters to fool users about which named entity they are communicating with.
For example, it might seem that
@user:matrix.org
and
@usеr:matrix.org
are identical; however, in the second instance the latin character e
in user
has been replaced by Cyrillic Ye, which is a homograph of e
in most Unicode fonts. Therefore, a malicious entity might trick others into thinking they are talking to user
when in reality they are not.
Unicode Technical Standard 39 describes an algorithm for detecting such "confusables". Fractal could apply this algorithm to the set of users in the current room and display a warning if any of their names are "confusable".
To my knowledge, no chat client on any platform currently implements such a feature, although similar algorithms are implemented in some web browsers.