Twitter shared an analysis which concluded that the reach of toxic content on the platform was lower than its own estimates
On Wednesday, the official Twitter Safety account shared the results of an independent assessment on the reach of toxic content on the platform.
Twitter Safety partnered with Sprinklr, a unified customer experience management platform, to conduct the analysis using an AI-powered model.
The analysis covered every English tweet publicly published between January and February 2023 and included a list of 300 English-language slur words.
Sprinklr's analysis found around 550,000 tweets containing at least one word from the list
The analysis did not include findings related to hateful images or symbols, and Sprinklr's definition of hate speech was narrower than Twitter's.
Twitter Safety explained that its focal metric is hate speech impressions rather than the number of tweets containing slurs.
The majority of slur usage is not hate speech, but Twitter works to reduce the reach of hateful content when it is identified.
According to Sprinklr's analysis, hate speech receives 67% fewer impressions per tweet than non-toxic slur tweets.Celebrities Gigi Hadid, Zayn Malik, Selena Gomez, and Leonardo DiCaprio