Twitter slow to remove racist and sexist tweets targeting Vice President Kamala Harris, report finds

The replies to tweets from Vice President Kamala Harris, the nation’s highest ranking Black female official, routinely include racist, sexist and graphic messages threatening her safety. A new report finds that hateful tweets directed at Harris are rampant, while Twitter has been slow to respond, and at times failed to act properly. 

According to Bot Sentinel, a non-partisan research organization that works to combat disinformation and targeted harassment online, during the first five months of this year, more than 4,200 tweets directed at Harris included the use of the n-word, manipulated graphic images and death threats. 

“All of those tweets violated their rules and policy when you look at them, and we made that available in the report,” Christopher Bouzy, the founder of Bot Sentinel, told CBS News. “It shows manipulated photos, for example, of her in lewd sex acts. I mean, that’s a no-brainer right there, but it came back that it did not violate their policies.”

A spokesperson for Twitter told CBS News the platform’s initial assessment of the tweets reported to them was incorrect and admitted mistakes were made in enforcing the platform’s content moderation rules.

Bouzy said his team began the research project after a group of preeminent Black women flagged to Bot Sentinel in January that Twitter was ignoring their numerous reports of abusive tweets directed at the vice president. 

“They were receiving the same messages that we ended up receiving, that those tweets were not violating their policies,” Bouzy said. He added he was “floored” by Twitter’s initial response to his reports about flagrantly racist and threatening tweets. 

Bot Sentinel, which uses Twitter’s engagement rules as a guide for its research studies, released a set of 40 tweets last week. Twitter initially removed only two of those, ignored 18 of them, and said the remaining 20 abusive tweets – including one threatening to kill Harris – didn’t violate the platform’s policies, according to Bot Sentinel. 

Bouzy said days after the initial report, Twitter suspended 10 of the accounts and removed all 40 tweets. A Twitter spokesperson said prior to Bot Sentinel’s report, 10 of the tweets referenced were reviewed by the platform but not found to be in violation of engagement rules. 

After further review, Twitter determined those tweets needed to be removed and that its initial assessment was incorrect. 

Twitter’s rules of engagement prohibit users from targeting others with repeated racial slurs that aim to dehumanize and promote negative stereotypes. The platform also forbids content that wishes or expresses a desire for the death of another user.

Penalties include requiring removal of tweets and suspending accounts when the platform determines their primary intent is to engage in hateful conduct. In some cases, Twitter will also downrank the tweets in the replies and make them ineligible for amplification in search results. The rationale here is to lower the risk of potential real world harm by limiting the visibility of the tweet on the platform. 

Twitter said the goal is to remove problematic tweets as fast as possible, before users even report it, but acknowledged mistakes happen when its content moderation teams are working on a large scale.

On Monday, Bot Sentinel released the second portion of its report, with screenshots and links to more than 1,300 tweets of a similar nature, in addition to the original 40 released last week. 

The research showing racist tweets directed at Harris comes as sentiments of white supremacy seep into American society and politics. Just last month, an 18-year-old drove over 200 miles, and killed 10 Black shoppers and injured three others, at a supermarket in Buffalo. Police said the shooting suspect was racially motivated, targeted the area because of its high percentage of Black residents, and wanted to kill as many Black people as he could. 

Several Republican candidates running for office in 2022 are looking to win support by warning voters of the “great replacement theory,” a racist talking point that falsely claims minorities are plotting with Democrats to diminish the role of White people in America.   

Online harassment of women from across the political spectrum has been documented on Twitter in the past. Politicians, journalists and media organizations are often the target of vitriol from users hiding behind a screen. 

But studies show women of color are three times more likely to be mentioned in abusive or problematic tweets, according to a report from Amnesty International. The 2018 study also found that Black women were 84% more likely than white women to be mentioned in hateful tweets. 

Bouzy said former Vice President Mike Pence was also routinely attacked by critics on Twitter, but the language directed at him wasn’t the same. He said Bot Sentinel has been reviewing tweets directed at Republican Rep. Marjorie Taylor Greene and Lauren Boebert, two representatives on the far right who are often at the center of controversy. 

“We wanted to see if it is similar, and they do face attacks as well,” Bouzy said. However, he added that it doesn’t compare to what Harris is subjected to because the sexist language and blatant use of racist remarks directed at the vice president “is on another level.”

The vice president’s office declined to comment on this story.

The barrage of hateful tweets that are being directed at Harris, and appear to break Twitter’s rules, could be a preview of what the platform might look like under Elon Musk’s leadership. Human rights advocates and social media experts have warned that Musk’s $44 billion bid to buy Twitter, and his promise to restore “freedom of speech” – so users can say whatever they want – will lead to proliferation of racist and sexist language on the platform. 

It is unclear if the problematic tweets targeting Harris are part of a coordinated campaign, and whether they come from real accounts or fake “bot” accounts which Musk has raised questions about. Many of the accounts appear to be real individuals with hundreds, and in some cases, thousands of followers. 

Tamer Hassan, co-founder and CEO of HUMAN, a cyber security firm that tracks fraudulent bot activity, says the content of the tweets gathered in the Bot Sentinel report are “not far from a typical” campaign organized by nefarious actors using bots.

“What we’ve seen over the last 10 years, the evolution of bot activity, is generally in the direction of one being very human-like,” Hassan said. “We typically see them in the social media world used for amplification.”

Hassan said organizing a campaign to direct hateful content towards someone like Harris “is very easy and low cost, virtually anybody can do it.”

“We are fast moving into a world where bots are a part of our culture and a part of our interactions,” Hassan said. 

“Bots are one of the most important threats to modern security, and there is very little policy or practice around their use,” he added. 

Tim Perry contributed to this report.