Hate speech on X now 50% higher under Elon Musk’s leadership, new study finds

Hate speech rose by 50 per cent on social media platform X in the months after billionaire Elon Musk bought it, according to a new study.

The study from the University of California Berkeley, which was published in the Public Library of Science journal on Wednesday, used artificial intelligence (AI) to randomly collect X posts that contained terms or phrases that are linked to English hate speech from Musk’s acquisition of X (then known as Twitter) in October 2022 until June 2023, when he stepped down as CEO. 

Researchers then sorted through them to check if each post met the criteria and to get rid of any duplicates or reposts. 

They also collected a “control group” of posts from January 2022 onwards to compare the amount of hate on the platform. 

The analysis found that hate speech doubled in the weeks before Musk’s Twitter takeover and lasted through till May 2023. 

During this time, the number of likes on posts that used specific homophobic, transphobic, and racist slurs went up by 70 per cent, suggesting that more users were exposed. 

“I think the increases in hate speech we see are concerning because that may mean… marginalised communities feel unwelcome or unsafe on the platform,” Dan Hickey, PhD student at Berkeley, told Euronews Next. 

The report continued that the number of bots and other inauthentic accounts did not decrease during this time despite Musk’s public pledges to reduce them. 

Their study follows others that have noted an increase in hate speech or antisemitism after Musk’s 2022 acquisition, includingreports from the Institute for Strategic Dialogue (ISD) and the Centre for Countering DigitalHate. 

More transparency needed from X on platform changes

Two spikes in hate speech happened before Musk became CEO: one in April 2022 to coincide with a widespread advertising campaign that featured a trans woman, the Berkeley report continued. 

The second was the few weeks in October ahead of Musk’s takeover. 

The report continued that a “reshuffling of leadership” happenedshortly after Musk became CEO, along with thedisbanding of the Trust and Safety Advisory Council that dealt with content moderation. 

However, the researchers say they can’t make “firm conclusions” about whether Musk caused the spike in hate speech on X because of limited information on “specific internal changes” on the platform. 

“To be able to really draw a definitive connection… one would need to know more about exactly what changed and when it changed, and the organisation has not been transparent about that,” Daniel Fessler, anthropology professor at Berkeley and part of the research team, told Euronews Next. 

After June 2023, Hickey said X started charging for veri collected by their Application Programming Interface (API) and they no longer had the budget to extend their study. 

What is X’s stance on hate speech?

Days after Twitter’s acquisition, Musksaid that the company’s “strong commitment to content moderation remains absolutely unchanged,” and that hate speech has declined since his ownership. 

“Freedom of speech doesn’t mean freedom of reach,” Musk wrote at thetime in a post on X. “Negativity should & will get less reach than positivity”.

X’s helppage says that users “may not directly attack other people on the basis of race,  ethnicity, national origin, caste, sexual orientation, gender, gender identity, religious affiliation, age, disability, or serious disease,” on the platform. 

The platform’s most recent transparencyreport disclosed that they suspended over 2,000 accounts and took down just under 5 million hateful posts of the more than 66 million hateful conduct user complaints filed between January and June 2024.

The company also counts hate speech in its “abuse and harassment” category. 

Of the 82 million complaints received, X suspended a million accounts and removed or labelled 2.6 million posts, the reportfound. 

More recently, Xsigned a voluntary EU pledge to fight online hate speech. The agreement asks social media companies to let experts monitor how companies review hate speech notices and teach users how to flag yasa dışı hate content on their sites.