Political violence is on the rise in the United States. According to a summary of key trends from the Princeton University Bridging Divides Initiative, this rise is reflected across a range of different statistics, from an increase in targeted violence and assassination attempts to an increase in the overall volume of threats and harassment against political figures at the local and national level.
Unprecedented levels of threats against public officials, including federal judges, both on- and offline have coincided with a bout of assassination attempts and acts of targeted violence in the United States. A growing number of violent acts over the past decade have demonstrated a clear nexus to the perpetrators’ social media use, including the September assassination attempt against President Donald Trump, Charlie Kirk’s murder, and the arson attack against Pennsylvania Governor Josh Shapiro.
We are now entering the midterm election cycle with more serious threats emanating from the online systems than ever before, and with fewer protections than we’ve had in a decade.
To be clear, perpetrators of violence are responsible for their own actions, and factors well beyond social media contribute to the increasing polarization in America. Nevertheless, online platforms have empowered once fringe and violent views to become more mainstream and are modern day tools for recruiting individuals; distributing messages intended to incite political violence; and organizing groups that may carry out violent acts. At the same time, online platforms have significantly degraded their policies and teams dedicated to trust and safety, as the federal government’s infrastructure for addressing foreign influence operations deteriorates.
Against this backdrop, we convened a group of experts—former tech company employees, policy makers, researchers, and academics—to consider what role online platforms are playing in potentially fueling violence and what the companies could and should do to mitigate these risks. We convened with a shared sense of urgency and recognition that the window for intervention is narrowing as polarization deepens, trust in institutions erodes, and acts of political violence become an increasingly normalized feature of American public life. We convened with an equal sense of urgency to chart a path forward that protects free expression and that fosters online spaces in which a genuinely free and fair competition of ideas can take place.
Today, we launched a report, titled “Political Violence in the Digital Age: What Online Platforms Can Do to Mitigate Escalating Threats.” Its 26 pages capture insights from the participants and include specific recommendations where we found consensus among the working group.
We are steadfast in our conclusion that these companies can, if they so choose, do more to help mitigate tech-fueled political violence. Accordingly, we are continuing to create a public record to counter the idea that—as executives often argue in the aftermath of real-world violence that had an online nexus—placing any significant responsibility on platforms is unfair.
Evolution of the online threat landscape and company actions
While societal ills such as political violence and polarization are not solely caused by social media, empirical research indicates a troubling relationship between structural features of online platforms—including their algorithmic promotion of certain types of content for engagement—and the severity of divisions in the United States. Evidence, including from Meta’s own internal research, suggests that the algorithmic targeting and amplification systems of large online platforms have at times led users—often young men and boys, who now perpetrate a large share of extremist and politically motivated violent acts—down radicalization pathways, helping violent narratives become more mainstream.
Extremist groups as well as individuals sowing division and encouraging violence utilize a widening range of both mainstream and fringe digital platforms and technological capabilities like AI to organize, communicate, and plan in a decentralized manner. More recently, we are seeing an increasing trend of tech-fueled violence associated not with organized groups with a clear ideological program, but rather self-initiated individuals influenced by harmful online communities and diffuse networks.
“These platforms serve as … the infrastructure through which divisive and violence-promoting narratives are distributed, amplified, and monetarily rewarded.”
These trends come as public officials–across the aisle–face growing threats and violent rhetoric online.
And yet, the major tech platforms have implemented what a February 2026 report from the Center for Democracy & Technology called “consequential changes” to their content moderation policies “that will influence the information environment ahead of the 2026 elections.” This includes substantial reductions in the trust and safety functions at the major platforms, with rolling layoffs of thousands of trust and safety workers in recent years.
Companies must step up to meet the moment
Platforms are unlikely to match the scale of their 2020 or 2024 preparations, but there are still steps they could take ahead of November. While it may seem futile to offer recommendations in this context, we believe it is important to continue providing potential interventions, policies, and design changes that the companies could employ. We do not anticipate that platforms will simply “do the right thing” without the proper regulatory and business incentives in place, but we aim to publicly document what companies could actually do, if they choose, to act more responsibly. The days where companies could say “we could not have seen it coming” are behind us.
Our list is not exhaustive and includes recommendations we and others have made before, but bear repeating. We prioritize concrete measures that companies could take to anticipate, monitor, and reduce the risks of political violence around the U.S. election, and how they might work with civil society to mitigate those risks.
The full set of nine recommendations—encompassing both immediate steps and broader design and product decisions required to reverse course toward a healthier and safer online environment—are spelled out in the report. Underpinning most of the recommendations is that platforms must prepare for the threat by investing in multidisciplinary expertise, robust threat assessment, and crisis response protocols for dangerous viral incidents. Other examples from our recommendations include:
- Clarifying and enforcing election and dangerous content policies: Platforms should publicly define and consistently enforce clear policies on election integrity, incitement to violence, harassment, AI-generated deception, foreign interference, and coordinated manipulation. This includes acting quickly against violent threats, temporarily slowing the spread of inflammatory viral content during crises, and improving cross-platform threat detection to identify emerging risks.
- Enforce rules equally, including against high-value users: Online platforms should uniformly apply and fairly enforce their existing rules for all users, regardless of status, political orientation, or follower count, including for influential figures whose content may disproportionately shape public discourse. Equally important, nobody, regardless of status or business importance to the companies, should be able to monetize content that incites violence.
- Protect targets of online threats: Platforms must be more responsive to user-generated reports, especially for those in imminent danger. Social media companies should, at the very least, make clear that they will not tolerate threats against public officials and election workers and put their money where their mouth is: resource the response.
- Stop incentivizing, profiting from, and rewarding divisive, inflammatory and dangerous behavior: Platforms should reduce algorithmic and financial incentives that amplify dangerous content, particularly during politically sensitive periods. This includes redesigning recommendation systems to promote healthier engagement and removing opportunities to monetize content linked to election denialism, hate speech, or glorification of violence.
On this last point: removing the ability to profit from election denialism, hate speech, glorification of violence, and potential incitement would diminish the financial and engagement incentives from those who use this type of content to grow their audiences and extract monetary value. This is a design choice: platforms can take steps to reduce or counterbalance these incentives.
While so much of the public conversation has moved on from social media to AI, it is urgent to remember that these platforms still serve as the megaphones through which political leaders and other influencers reach the public, and the infrastructure through which divisive and violence-promoting narratives are distributed, amplified, and monetarily rewarded.
With less than five months remaining before November’s elections, technology platforms have a crucial role to play in helping turn down the temperature, and they must be prepared in the event that political violence occurs. The recommendations we provide here should represent the floor, not the ceiling: minimum steps platforms can take in a broader commitment to an information environment that nurtures democracy rather than division and political violence.
