Is there a proactive way to defend against invisible Unicode attacks?

Filtering them?