May 11, 2021

Twitter announced a new feature that detects “mean” replies on its site before a user presses send. In recent months, users have urged the popular social media platform to do something about harassment and hateful messages on the platform.

The new feature, introduced on May. 5, 2021, will initially be activated on accounts with English-language settings, but it is unclear when it will be offered to accounts with other languages.

The new feature is able to detect a not-very-nice tweet and append an automatic prompt that says, “Want to verify this before tweeting?” The user is then presented with three options: tweet, edit, or delete.

Twitter said that after a year of testing the new feature with the add-on, it found that users sent fewer offensive responses through the service, with the social network reporting that 34% of users either modified their original response or decided not to send it when asked.

After their first prompt, users wrote about 11% fewer offensive replies in the future, reducing the chances that they would receive offensive replies in return.

For more information, read the original story in NPR.