It follows a number of footballers speaking out about experiencing racist, sexist and other abuse on Instagram.
Direct messages (DMs) containing words or emojis deemed offensive will be removed from view.
The tool will be available in the UK, France, Ireland, Germany, Australia, New Zealand and Canada within weeks.
More countries will then receive the relevant update in the coming months.
"Because DMs are private conversations, we don't proactively look for hate speech or bullying the same way we do elsewhere," Instagram blogged.