Twitter adds Arabic to ‘Reply Prompts’ feature to encourage people to rethink potentially harmful or offensive replies

Today, Twitter adds Arabic to “Reply Prompts’’ encouraging people to rethink potentially harmful or offensive replies - such as insults, strong language or hateful remarks before they hit send, in line with its commitment to create a space that encourages meaningful conversations.

People come to Twitter to talk about what's happening, and sometimes conversations can get intense and people say things in the moment they might regret later. Experiments demonstrate promising results with people changing or deleting the replies over 30% of the time when prompted for English users in the US and around 47% of the time for Portuguese users in Brazil.

Launched in 2022, globally in the English language along with Spanish in Mexico, Turkish globally and Portuguese in Brazil, Reply Prompts encouraged people to pause and reconsider a potentially harmful or offensive reply before they hit send. Based on feedback and learnings from those tests, the platform made improvements to the systems that decide when and how these reminders are sent. Starting today, the feature will be available on iOS, Android and Web for accounts that have enabled Arabic-language settings in Saudi Arabia.



Identifying the need

In early tests, people were sometimes prompted unnecessarily because the algorithms powering the prompts struggled to capture the nuances of many conversations and often didn't differentiate between potentially offensive language, sarcasm, and friendly banter. Throughout the experiment process, Twitter analyzed results, collected feedback from the public, and worked to address its errors, including detection inconsistencies.

These tests ultimately resulted in people sending less potentially offensive replies across the service, and improved behavior on Twitter. The platform learned that:

If prompted, 30% of English users in the U.S. and 47% of Portuguese users in Brazil revised their initial reply or decided to not send their reply at all.
After being prompted to reconsider their replies, users canceled them 9% of the time and revised them 22% of the time.
Those who were prompted posted 6% fewer offensive Tweets
If prompted, people were less likely to receive offensive and harmful replies back.

Since the early tests, here’s what Twitter’s incorporated into the systems that decide when and how to send these reminders:

Consideration of the nature of the relationship between the author and replier, including how often they interact. For example, if two accounts follow and reply to each other often, there’s a higher likelihood that they have a better understanding of preferred tone of communication.
Adjustments to Twitter’s technology to better account for situations in which language may be reclaimed by underrepresented communities and used in non-harmful ways.
Improvement to the platform’s technology to more accurately detect strong language, including profanity.
Creation of an easier way for people to let them know if they found the prompt helpful or relevant.


What’s next

Twitter will continue to explore how prompts — such as reply prompts and article prompts — and other forms of intervention can encourage healthier conversations on the platform. It will also continue to collect feedback from people who have received reply prompts as the platform expands this feature to other languages and markets.

Stay tuned for more updates as Twitter continues to learn and make new improvements to encourage more meaningful conversations on the platform.