Twitter on Wednesday announced it's released a feature that detects
Caption

Twitter on Wednesday announced it's released a feature that detects "mean" replies on its service before a user presses send. / AP

Twitter wants users to think twice about sending a mean or offensive tweet.

The tech company on Wednesday announced it has released a feature that detects "mean" replies on its service before a user presses send. When a not-very-nice tweet is detected, an automatic prompt reads, "Want to review this before Tweeting?" The user is presented with three choices: tweet, edit, or delete.

This feature, which launched Wednesday, will initially be enabled on accounts with English-language settings. It's unclear when other languages will be added.

Twitter says after a year of tests with the add-on, users were sending fewer offensive replies across the service.

According to Twitter, when prompted, 34% of people revised their initial reply or decided not to send it at all. After they were first prompted, users composed about 11% fewer offensive replies in the future. They were also less likely to receive offensive replies in return.

Users, especially women, receive a high level of harassment and abuse on the service, research shows.

In 2018, Amnesty International analyzed 228,000 tweets sent to 778 female politicians and journalists in the U.K. and U.S. in 2017. The subsequent report revealed 1.1 million "abusive or problematic tweets" were sent to women throughout the year — an average of nearly one per 30 seconds.

The problem is worse for women of color, the organization found. Black women are 84% more likely than white women to be targets of abusive tweets.

High-profile Twitter users have regularly called on the company to do something about harassment and hateful messages. Rep. Alexandria Ocasio-Cortez has spoken publicly about the level of harassment and death threats she receives on Twitter.

Twitter's move this week is just the latest by a major social media company to take steps to modify users' behavior. Instagram is testing ways to hide "like" counts on posts in an effort to reduce social media addiction, jealousy, anger and depression.

Twitter said its new feature takes into account the context of conversations — making exceptions for offensive language used sarcastically or as part of "friendly banter." It also weighs the nature of the relationship between the message's author and the person replying, including how often the two interact.

Copyright 2021 NPR. To see more, visit https://www.npr.org.