Midterms highlight rise in hate speech
asserted
Midterms → highlight → speech
-
- After Democrats’ midterm victory, Facebook posts called for armed conflict against Muslims and socialists, underscoring a surge in violent political rhetoric online.
-
asserted
posts → call → rhetoric
Researchers found threats against candidates rose fivefold on X before the midterms; a separate analysis found that violent and hateful posts targeting politicians quadrupled on Facebook after Meta cut fact-checking.
asserted
Meta → find → checking
- Experts say looser moderation has amplified hate and misinformation, while calls grow for consistent platform rules, transparency and stronger regulation.
asserted
calls → say → rules
A day after the Democrats swept to victory in the midterms, online right-wing discourse in the aging corners of Facebook turned to calls for an armed conflict with Muslims and socialists.
asserted
discourse → sweep → Muslims
“Vote from the rooftops” said the caption on an image of a sniper taking aim from a rooftop.
asserted
caption → vote → rooftop
The slogan is a euphemism in rightwing subculture that if voting fails, it’s time to take up arms.
asserted
it → fail → arms
Another post warned of a coming civil war with “no one left to protect Muslims and social democrats” if they defund the police.
asserted
they → warn → police
“I got a feeling that Abdul El Sayid won’t be around too long,” said another post, referring to the Michigan senator-elect, who faced one of the ugliest islamophobic campaigns.
asserted
who → get → campaigns
But the problem has grown worse in recent years, experts said, after social media platforms including Meta-owned Facebook and Instagram rolled back content moderation and fact-checking policies over fears of censorship and bias.
asserted
platforms → grow → censorship
Anti-immigration rhetoric, election denialism and health skepticism are recurring themes for violent and false narratives in extreme right-wing online subcultures, weaponized by politicians.
asserted
rhetoric → recur → politicians
Incendiary speech has been further boosted by AI-powered slop that’s worsened an ailing information ecosystem.
asserted
that → boost → ecosystem
In the runup to the midterms, wishes of harm and violent threats targeted at candidates running for office jumped fivefold on Elon Musk’s X, according to the Center for Countering Digital Harm, a non-profit based in Washington DC.
uncertain
wishes → target → DC
The research analysed replies to posts from ten candidates in battleground races from Michigan to Texas, and found hundreds calling for violence.
asserted
hundreds → analyse → violence
A similar analysis in June found violent threats and hate against U.S. politicians on Facebook quadrupled after Meta cut its fact checking program in January 2025, according to CCDH.
uncertain
Meta → find → CCDH
Hate and violent speech targeted at President Trump also doubled during that period.
asserted
Hate → target → period
“These findings point to a continued pattern of platforms’ safeguards failing to proactively remove content that contributes to our increasingly violent political culture,” said Imran Ahmed, chief executive of CCDH, a nonprofit based in Washington.
asserted
Ahmed → point → Washington
A spokesperson for Meta said the company regularly issues “public reports tracking violating content on our platforms, and the prevalence of hateful conduct did not increase throughout 2025.”
asserted
prevalence → say → 2025
Meta removed content it found to violate its policies, the spokesperson said.
asserted
spokesperson → remove → policies
Platforms have long struggled to strike a balance between free expression and censorship.
asserted
Platforms → struggle → expression
Meta’s Mark Zuckerberg began as a free speech absolutist, but had to beef up moderation and introduce third party fact checking after anti-minority hate speech on Facebook fueled real world violence in Myanmar, Sri Lanka, and Ethiopia a decade ago.
asserted
speech → begin → Myanmar
The debate intensified during the COVID 19 pandemic over censoring speech that doubted or joked about the origins of the virus and anti-vaccine content, and culminated in the removal of Donald Trump’s account after inflammatory posts on the day of Jan. 6, 2021, Capitol riot.
Elon Musk’s takeover of Twitter in 2022 changed the moderation landscape even further.
asserted
takeover → intensify → landscape
As the billionaire railed against “woke” culture, he fired the platform’s moderation and fact-checking teams.
asserted
he → rail → teams
Musk replaced it with the crowdsource program known as Community Notes that enables users to fact-check and flag problematic posts.
asserted
that → replace → posts
When Trump returned as President in 2024, the pendulum had swung back.
asserted
pendulum → return → 2024
Accused of censoring conservative views, Zuckerberg in public hearings attributed the pandemic-era speech policing to government pressure, and expressed regret over not speaking sooner.
asserted
Zuckerberg → accuse → regret
In 2025, he said that across Instagram, Facebook and Threads, the company would remove content filters around immigration and gender.
asserted
company → say → immigration
It means we’re going to catch less bad stuff, but we’ll also reduce the number of innocent people’s posts and accounts that we accidentally take down,” Zuckerberg said in a video message.
asserted
Zuckerberg → mean → message
He described Trump’s 2024 victory as a “cultural tipping point” to correct the over-censorship, and replaced fact checking with Community Notes.
asserted
He → describe → Notes
Now, explicit calls for civil war or posts calling to prepare for the coming fight are on the rise on social media, experts say, especially in the lead up to and days after the midterms.
asserted
experts → call → midterms
“So are we wearing camo or what?
asserted
we → wear → camo
Sorry, this will be my first ever civil war,” read a Facebook post.
asserted
post → read → ?
Another image with a rifle read: “If it’s time to hide them, then it’s time to use them.”
asserted
it → read → them
“Problematic content on Meta platforms, which is primarily Facebook and Instagram, has increased after Meta rolled back its kind of fact-checking and anti-hate speech policies,” said Magdalena Wojcieszak, UC Davis professor of communication.
asserted
Wojcieszak → increase → communication
She pointed to an April study that found that Instagram failed to remove 93% of flagged hateful and extremist content by white supremacist and Nazi accounts.
asserted
Instagram → point → accounts
Wojcieszak’s research showed that Meta’s stricter moderation before the 2020 election reduced misinformation exposure by 72%.
asserted
moderation → show → %
However, contrary to popular belief, reducing exposure alone did not alter political attitudes or diminish polarization.
asserted
reducing → reduce → polarization
Meta has argued that the greater tolerance of offensive speech has resulted in fewer overall complaints about content takedowns.
asserted
tolerance → argue → takedowns
“Should we prefer to catch 95% of hateful speech at the cost of mistakenly flagging 1% of unobjectionable content, or is it better to live in a world in which 99% of hateful speech is caught but 5% of unobjectionable content is taken down?
asserted
% → prefer → content
There’s no right answer,” said Olivier Boutin, UC Berkeley PhD student, studying political misinformation.
asserted
Boutin → ’ → misinformation
…and 15 more, not listed.
The Aporia
