AI xPress

AI and computer science research.

Back to Foragentis

AI & Society

The weekly rate of hate speech on X stayed about 50 percent higher than in the months before Elon Musk bought the platform through at least May 2023, and activity by inauthentic accounts did not decline after the purchase

AI & Society November 17, 2026

Canvases and brushes in an artist's studio

In a 2025 study published in PLoS ONE, researchers examined material posted on X, the platform formerly called Twitter, from the start of 2022 through June 2023. They found that the weekly rate of hate speech rose just before Elon Musk bought the platform and stayed about 50 percent higher than in the months before the purchase through at least May 2023. Likes on hate posts doubled, which the researchers describe as a sign of increased engagement with hate posts. The increase appeared across racism, homophobia, and transphobia. The researchers say they cannot directly tie the rise to any policy at X. The National Science Foundation, the Defense Sciences Office, and DARPA funded the work.

Did X's policies cause the rise in hate speech?

The researchers could not directly tie the hate-speech increase to any policy at X. The observed rise had begun just before the purchase and persisted, but the study describes the pattern without assigning that cause.

What did the study find about inauthentic accounts?

Activity by inauthentic accounts did not decline after Musk bought X, and it may have increased. These accounts are often used in spam and malicious information campaigns. The authors say that this could point to further negative outcomes, such as the potential for scams, interference in elections, or harm to public health campaigns.

What period did the study cover?

The researchers examined material posted on X from the beginning of 2022 through June 2023, a window that includes Musk's full tenure as CEO. The hate-speech increase had already appeared just before the purchase and lasted until at least May 2023.

Limits and what's next

The study examined material posted on X from the start of 2022 through June 2023. The abstract says the hate-speech increase cannot be directly attributed to any policy at X. The abstract does not say whether the findings were independently replicated. The findings describe one platform in one period, and the abstract does not say whether the patterns continued after June 2023 or would appear on another platform.

Source: Daniel Hickey et al., “X under Musk’s leadership: Substantial hate and no reduction in inauthentic activity,” PLoS ONE, 2025.