Twitter announced Tuesday a new policy to combat “dehumanizing language,” as it continues searching for ways to make its platform more hospitable.

Users will not be allowed to “dehumanize anyone” based on their background because “this speech can lead to offline harm,” the company said in a blog post. If you’re wondering at this point what  Twitter’s definition of “dehumanizing language” is, the company broke it down into two categories: when users are “denied of human qualities” or “denied of human nature.” Twitter gave examples of comparing users to animals or “reducing groups to their genitalia.”

Twitter said allowing these insults can be a driving force behind “normalizing serious violence,” and pointed to a few scholars to bolster its stance. Before Twitter’s new policy goes into effect on Oct. 9, users will be able to submit their thoughts on the rules and if they can be improved. The questionnaire is at the bottom of the company’s blog post.

Also Read: Twitter 'Unintentionally Sent' Users' Direct Messages to Third-Party Developers

Tuesday’s announcement expands on Twitter’s established rules against “hateful conduct,” which bars users from attacking others for their racial, religious, or sexual makeup. The company said the “dehumanizing” policy is part of its 2018 mission to improve the “health of the public conversation” on its platform — something CEO Jack Dorsey reiterated to Congress a few weeks ago. Some measures have already been adopted, like hiding mean tweets at the bottom of reply threads.

The update comes as Twitter has recently been championed and criticized for its rules enforcement, after digital conspiracy theorist was permanently suspended for violating its “abusive behavior” policy earlier this month.