Twitter spaces filled with hate speech and extremism, report finds

Twitter began testing its live audio chat feature in November 2020.

James Martin / CNET

Taliban supporters, white nationalists and anti-vaccine activists have reportedly flocked to Twitter Spaces’ live audio chat feature, the social media site struggling to control harmful content.

The Washington Post, citing researchers and users, as well as screenshots seen by the outlet, reported on Friday that hundreds of people had listened to the offensive live audio conversations. Twitter users made derogatory comments about Shia Muslims, belittled transgender and black Americans, and spread disinformation on Spaces, according to the Post.

Twitter allows users to report live audio chats that violate its rules, and the company records a copy of a flagged audio chat for at least 30 days to investigate rule violations. Twitter also uses software to detect offensive keywords in Space titles. The social media site, however, does not have human moderators or technology that analyzes audio in real time. Current and former employees told the Post that Twitter has also felt pressure from investors to post products before they are tested for safety. Twitter co-founder Jack Dorsey stepped down as CEO in November and Parag Agrawal, who was the company’s chief technology officer, stepped in to fill the role.

Twitter started testing Spaces in November 2020 as a social audio app Club house gained popularity. Clubhouse has also struggled to moderate misinformation, especially on COVID-19 vaccines.

Twitter’s software also mistakenly promoted some of the harmful live audio chats, which the company attributed to a “bug,” the Post reported.

A Twitter spokeswoman told CNET that over the past week the company discovered “bugs” in its software and that this increased the time it takes to delete audio chats that broke its rules. “This left harmful conversations in the foreground and was a mistake. We have fixed these bugs and will continue to explore ways to further extend our proactive detection, as well as to evaluate and develop new moderation options,” she declared.

CNET asked Twitter how many audio chats it had deleted for breaking its rules, but the company did not provide that information.

Comments are closed.