The company originally started testing the filter in August and has now implemented it across the platform.
Twitter users will now have an "additional messages" folder under "Message requests" in their inbox.
Should a direct message contain what appears to be suspicious language, spam, or offensive content, it will be diverted to this folder.
A user will then have to go into the "Message requests" section and tap on "Show additional messages" to view it.
The change means that Twitter users can still interact with people they don't follow without leaving their inboxes exposed to abuse.
In an update delivered earlier this year, Twitter introduced reply moderation, which lets users hide specific responses to their tweets.
The efforts come as Twitter is trying to clamp down on widespread harassment on its social media platform.