Twitter introduces Safety Mode, which blocks harassing accounts for seven days, in beta for some English-language accounts
Feeling safe on Twitter looks different for everyone. We've rolled out features and settings that may help you to feel more comfortable and in control of your experience …
TwitterJarrod Doherty
Context & Ripple Effects
Safety Mode is the productization of a plan Twitter previewed six months earlier, when it said it was building a mode that would automatically block and mute accounts acting abusive or spammy. The seven-day block lands at the end of a long escalation ladder: automatic filtering of abusive mentions back in 2015, then hiding abusive tweets and safe search in early 2017.
What distinguishes this step is that it shifts enforcement from manual tools — the word-muting and account-muting controls of 2017 — to automated action taken on the user's behalf, with the platform deciding which replies qualify as harassment.
First-order effects
Users in the beta get reply-level protection without lifting a finger: accounts Twitter flags as harmful are blocked for seven days, replacing the mute-and-block choreography users previously had to perform manually.
Accounts flagged by the automation face temporary exclusion from conversations they did not consent to, with no human review visible in the rollout — accuracy of Twitter's classifier now directly determines who can participate.
Second-order effects
If the beta performs, Twitter has a template for scaling enforcement beyond its moderation staff, as shown when it later widened the test to half of US users plus five other markets (the February 2022 expansion) — competitors face pressure to match automated per-user defenses rather than one-size-fits-all policy takedowns.
Third-order effects
The pattern points toward platforms delegating routine abuse handling to algorithmic, time-boxed penalties that sit between muting and permanent suspension — a middle layer of enforcement whose error rates and appeal paths become the real regulatory question.
Automated blocking also raises the stakes of false positives: if classifiers silently silence legitimate participants, trust in conversation quality becomes a measurable product differentiator across social networks.
The trend: Social platforms are moving from user-operated moderation tools toward automated, time-boxed enforcement actions taken on users' behalf, with Safety Mode marking Twitter's shift from filters that hide abuse to blocks that punish it.
Now testing: Safety Mode to help reduce disruptive interactions on Twitter. Automatically block accounts that add unwelcome replies, Quote Tweets, and mentions to your convos. If you're in the test, you can turn on Safety Mode in your “Privacy and safety” settings. https://twitte…
we're rolling out “safety mode” as an experiment to help people automatically block and tune-out potentially unwelcome interactions more easily. https://twitter.com/...
1984 is a great fiction novel to read but it seems like it is becoming the reality we are currently living under more and more each day https://twitter.com/...
Why can't we just ban abusive accounts????? Oh right because numbers & shares and ad money is more important than real users mental health and physical well being... https://twitter.com/...
Looks like some shit for losers & kindergardaners. Do not add any new features unless theyre good or they put 1000usd directly in my pocket. https://twitter.com/...
What about taking away the ability of locked/private accounts to QT accounts that are not following the locked account? It would be a much more practical and popular feature https://twitter.com/...
So instead of banning these people when you report them, they are initiating a ‘safety mode’ auto-block feature that nobody will use because it will probably also auto-block friends making jokes in your comments. Cool. https://twitter.com/...
Oh cool so they made a feature that will help journalists with trash takes who are getting ratioed. *But still didn't fix the child porn problem. Twitter is testing a new anti-abuse feature called ‘Safety Mode’ | TechCrunch https://techcrunch.com/...
Finally. I have trolls that quote tweet me even though they're mutually blocked. 🤷🏾♂ ️ Hopefully this will stop their pathological nonsense. Thanks, @TwitterSupport! https://twitter.com/...
“Author of Tweets found by our technology to be harmful or uninvited will be autoblocked, meaning they'll temporarily be unable to follow your account, see your Tweets, or send you Direct Messages.” https://twitter.com/...
We'll be testing Safety Mode with a small group of people on Twitter in the coming months and expanding the pool of beta testers, as we receive feedback. Stay tuned for more updates and head to our blog to learn more about Safety Mode. https://blog.twitter.com/...