Twitter is moving to filter out inappropriate content based on religion as part of its effort to curb hate speech.
In a policy update on Tuesday, Twitter said it would take down "dehumanising language" that targets specific religious groups.
Examples of the type of language that would be removed would be the description of a user's religion as "disgusting" or referring to them as "filthy animals".
"After months of conversations and feedback from the public, external experts and our own teams, we're expanding our rules against hateful conduct to include language that dehumanises others on the basis of religion," the Twitter safety team wrote.
"Starting today, we will require tweets like these to be removed from Twitter when they're reported to us."
Tweets that break this rule sent before the update will need to be deleted, but will not result in account suspensions, according to the company.
Twitter and other social networks have been struggling with ways to remove hateful and abusive content while remaining open platforms.
The online services have also faced accusations that the curbs end up suppressing some political views.
Twitter last month said it would add warning labels to tweets from officials and politicians that violate its rules - a move potentially affecting the prodigious output of US President Donald Trump.
Trump is holding a "social media summit" this week, inviting conservative critics of Big Tech but excluding the big online platforms themselves, where some of the complaints are likely to be aired.