Twitter Inc announced on Wednesday that it will roll out a safety feature that would allow users to temporarily block accounts for seven days if they use abusive language or send unsolicited responses.

When Safety Mode is enabled, Twitter’s computers examine the content of tweets to determine the likelihood of a negative engagement and the author-replicator connection.

Accounts that are regularly interacted with will not be automatically blocked, according to the firm, since it considers existing relationships.

Twitter has earlier taken several steps to address harassment on its site, which often occurs in unsolicited replies targeting women and minorities.

“We want people on Twitter to enjoy healthy conversations, so we’re limiting overwhelming and unwelcome interactions that can interrupt those conversations,” the company said.

Safety Mode can be turned on under settings and will be available to a small feedback group on iOS, Android and Twitter.com, beginning with accounts that have English-language settings enabled, Twitter said.