Violent threats against candidates rise 500% on X

One day after Democrats secured a sweeping victory in the midterm elections, online discourse in right-wing Facebook groups shifted toward calls for armed conflict. Posts circulating on the platform demanded action against Muslims and socialists. One image featured a sniper aiming from a rooftop with the caption, “Vote from the rooftops,” a phrase used in subculture to suggest taking up arms if voting fails.

Researchers at the Center for Countering Digital Harm, a Washington DC-based nonprofit, tracked this escalation. Their data shows that wishes of harm and violent threats targeting candidates jumped fivefold on X in the runup to the elections. The analysis covered replies to posts from ten candidates in battleground races ranging from Michigan to Texas. The team identified hundreds of posts calling for violence.

A separate review by the Center for Countering Digital Harm found similar trends on Meta-owned platforms. Violent threats and hate speech directed at U.S. politicians on Facebook quadrupled after Meta ended its fact-checking program in January 2025. Hate and violent speech targeting President Trump doubled during that same period.

Imran Ahmed, chief executive of the Center for Countering Digital Harm, stated that platform safeguards are failing to remove content that contributes to a violent political culture. He emphasized that the findings point to a continued pattern of inaction.

A Meta spokesperson responded by noting that the company regularly issues public reports tracking violating content. The spokesperson said the prevalence of hateful conduct did not increase throughout 2025. Meta removed content it determined violated its policies.

Magdalena Wojcieszak, a UC Davis professor of communication, said problematic content on Meta platforms has increased since the company rolled back fact-checking and anti-hate speech policies. Her April study found Instagram failed to remove 93% of flagged hateful content from white supremacist and Nazi accounts. Earlier research by Wojcieszak showed that stricter moderation before the 2020 election reduced misinformation exposure by 72%.

Platform leaders have framed the changes as a necessary trade-off. Mark Zuckerberg explained in a video message that looser moderation means catching less bad content but reducing accidental takedowns of innocent posts. He described the 2024 election result as a cultural tipping point that corrected over-censorship. Meta replaced its third-party fact-checking with the Community Notes program.

Olivier Boutin, a PhD student at UC Berkeley studying political misinformation, questioned the balance between catching hateful speech and preserving unobjectionable content. He asked whether it is better to catch 95% of hateful speech at the cost of flagging 1% of good content, or to catch 99% of hate at the cost of taking down 5% of good posts. He said there is no right answer.

Wojcieszak said lasting change requires regulation and penalties for platforms spreading hate. She noted that the Democratic win alone will not change moderation policies. Alex Karp, chief executive of Palantir, reasserted his progressive credentials this week. Joel Kaplan, who helped connect Zuckerberg to the Trump administration, will leave Meta after 15 years.

Ahmed of the Center for Countering Digital Harm said people want transparency and accountability. He argued that if platforms contribute to real-world harm, they should be held liable. He criticized the current approach as a ridiculous pendulum swinging between strict moderation and free speech based on who is in power.