Twitter says fewer people are reporting abusive posts on their service thanks in part to a combination of machine learning and proactive customer service that acts on detectable terms of service violations.
In a letter to shareholders released on Monday, Twitter said user reports of abusive tweets was down 27 percent in the final quarter of 2019 compared to the previous quarter. It did not disclose the exact number of abuse reports it received in either quarter.
Twitter said its advanced machine learning technology helped send potentially-abusive posts to content moderators for review, which resulted in less users having to report abusive tweets on their own. The company said implementing technology that helped the service catch problematic tweets earlier was part of its overall strategy to improve health and wellness on the platform, which remains a priority through 2020.
Early last year, Twitter relaunched its “rules” page, which outlines the content Twitter prohibits on the service. It clarified that tweets that contain glorifications of violence, targeted harassment or promotion of violent extremism and terrorism were not welcome on the platform.
It also said posts that contain private information, such as home addresses and phone numbers not publicly available online, cannot be posted without “express authorization and permission.”
That term has caused some confusion among professionals who use the service, particularly those who publish investigative journalism. Earlier this month, reporter Scott Stedman with Forensic News was prevented from accessing his account over a January tweet that supplemented his reporting on ties between Deutsche Bank and a Russia-backed financial firm.
The tweet contained a screen capture of a Microsoft Excel document showing funds flowing into Deutsche Bank’s American subsidiary, including a sizable amount from Russia’s Gazprombank. The timing was newsworthy, Stedman reported, because it coincided with a significant loan executed by then-businessman Donald Trump in 2013.
Speaking on background, a Twitter employee told The Desk Stedman’s tweet contained “numbers” that violated the company’s terms of service. Stedman, who co-founded Forensic News, agreed to delete the tweet in order to regain access to his account; in a follow-up post, he said he rejected Twitter’s position regarding the tweet.
“I ended up deleting the tweet in question because I need my account to support my journalism and business,” Stedman wrote shortly after his restriction was removed.
