Twitter is broadening access to a feature called Safety Mode, designed to give users a set of tools to defend themselves against the toxicity and abuse that is still far too often a problem on its platform. First introduced to a small group of testers last September, Safety Mode will today launch into beta for more users across English-speaking markets including the U.S., U.K., Canada, Australia, Ireland, and New Zealand.
The company says the expanded access will allow it to collect more insights into how well Safety Mode works and learn what sort of improvements still need to be made. Alongside the rollout, Safety Mode will also prompt users when they may need to enable it, Twitter notes.
As a public social platform, Twitter faces a continual struggle with conversation health. Over the years, it’s rolled out a number of tweaks and updates in an attempt to address this issue — including features that would automatically hide unpleasant and insulting replies behind an extra click; allow users to limit who could reply to their tweets; let users hide themselves from search; and warn users about conversations that are starting to go off the rails, among other things.
But Safety Mode is more of a defensive tool than one designed to proactively nudge conversations in the right direction.
It works by automatically blocking accounts for 7 days which are replying to the original poster with harmful language or sending uninvited, repetitive replies — like insults and hateful remarks or mentions. During the time that Safety Mode is enabled, those blocked accounts will be unable to follow the original poster’s Twitter account, see their tweets and replies, or send them Direct Messages.
Twitter’s algorithms determine which accounts to temporarily block by assessing the language used in the replies and examining the relationship between the tweet’s author and those replying. If the poster follows the replier or interacts with the replier frequently, for example, the account won’t be blocked.
The idea with Safety Mode is to give users under attack a way to quickly put up a defensive system without having…