Making sense of rumors about the Trump assassination attempt

2024 U.S. ELECTIONS RAPID RESEARCH BLOG

By Danielle Lee Tomson, Kate Starbird, Nina Lutz, Stephen Prochaska, Ashlyn B. Aske, Melinda McClure Haughey, Joseph S. Schafer, Zarine Kharazian, Adiza Awwal, and Michael Grass
Center for an Informed Public
University of Washington

This is part of an ongoing series of rapid research blog posts and rapid research analysis about the 2024 U.S. elections from the University of Washington’s Center for an Informed Public.

Key Takeaways

• After the assassination attempt on presidential candidate Donald Trump, people converged online to make sense of available information in a process known as collective sensemaking. 
• Collective sensemaking has two core components: assembling potential evidence and framing. The available evidence helps us select the frames we use, and the frames shape which evidence we focus on and how we interpret it.
• Early sensemaking focused on determining who the shooter was and why he had done it. We expect continued speculation online about the shooter’s motives, along with selective evidence sharing and continued framing contests.
• We also observed three politically-coded frames emerge: On one anti-Trump side, conspiratorial framing that the assassination attempt was “staged” for political gain; on the pro-Trump side, equally conspiratorial framing of the assassination attempt as an “inside job;” and in between, framing that criticized the Secret Service for failing to adequately protect the former President. 
• Savvy creators are also utilizing AI to make content, memes, and product listings in response to the attempted assassination, including AI-generated imagery and commemorative merchandise listings. 

On Saturday afternoon (July 13, 2024), a campaign rally stage in rural Butler, Pennsylvania became the site of a horrific and tragic event: an assassination attempt on former president Donald J. Trump. The presumptive 2024 Republican presidential nominee had been speaking during the rally when audience members reported hearing shots fired. Within seconds, members of Trump’s security detail surrounded him as he took cover behind the podium. Moments later, he would emerge from the scrum with his face bloodied and his fist raised, captured in immediately iconic photographs that lit up the internet.

The aftermath of the event was chaotic as officials, rally attendees, and online audiences tried to make sense of a dynamic and, at times, conflicting information space. Initial media reports containing hedgy language about “popping noises” were soon updated to assert that shots had indeed been fired. News spread that three audience members had been hit by gunfire, which resulted in one fatality. Photos circulated of a suspected shooter who lay dead on a rooftop a few hundred feet away. As the facts crystallized, information participants worked to determine the frames through which those facts would be interpreted. Diverging along ideological lines, pro-Trump rhetoric attempted to assign blame for the shooting, anchoring on an assumption that the perpetrator was a Democrat motivated by political rhetoric acutely critical of Trump. Meanwhile, among anti-Trump commentators, a sense of skepticism spread, with many theorizing that the event may have been “staged” by the Trump campaign for political gain. 

For researchers familiar with how information flows during crisis events (and specifically in the aftermath of mass shootings) these dynamics — unpredictable and uncertain information spaces, political framing contests, conspiracy theorizing — are not surprising. Instead, they are reflective of core characteristics of many crisis events, the sociotechnical structure of our modern information spaces, and our political moment.

In this article, we share some of what we have seen on social media and other online spaces in the aftermath of the Trump assassination attempt, and put that into context based on what we know about how people use social media during crises and breaking news events. Specifically, we map some of the emerging narratives that we are tracking to a “collective sensemaking” approach for understanding rumoring and conspiracy theorizing during and after crisis events. Throughout the piece and drawing on our longstanding expertise of collective sense-making dynamics, we share some thoughts about what we might anticipate in the coming days as this situation evolves. Our hope is that this article will be helpful to everyday people participating in this sensemaking process, as well as journalists attempting to cover these conversations.

Background: What Is Collective Sensemaking?

During crisis events like Saturday’s shooting, we often see groups of people come together to make sense of what is happening amid intense anxiety and informational ambiguity. This dynamic process, called “collective sensemaking,” is increasingly happening online. Rumoring — with rumors defined as stories or pieces of information unverified at the time of propagation — is a natural part of this process, as groups develop theories to explain information and evidence that emerges. According to sociologist Tamotsu Shibutani’s 1966 book, rumors help us assign meaning in otherwise uncertain, rapidly unfolding events. Seen from this perspective, rumors can be either true or false, given they are defined not by their facticity but rather their unverified nature and spread through informal channels. In contemporary life, rumoring has found a home on the internet and has become an integral part of digital life where instantaneity is increasingly valued. 

Drawing on the work of Gary Klein, et al, our approach to understanding how groups engage in collective sensemaking online focuses on the interactions between shifting evidence and frames, or ways to interpret this evidence. Groups adjust frames to suit multiple purposes or interests as new evidence emerges, such as first-hand social media photos and videos, reporting, witness testimony, past experiences, photojournalism, and expert opinion. Drawing on the same evidence, oppositional groups may contest framing of what is going on during and after the crisis. We have observed in prior research that emerging evidence may be boosted and/or reframed by larger, more influential accounts we call “news brokers,” who play an outsized role in the framing process. As we’ve written before, disinformation can be seen as manipulation of the sensemaking process (both through the introduction of faulty evidence and the intentional shaping of frames), and conspiracy theorizing occurs when these sensemaking processes go awry (often due to early anchoring on a conspiratorial frame).

The subsequent examples illustrate this dynamic relationship between evidence and frames in the process of collective sensemaking that we have seen in this assassination attempt crisis, informed by learnings from previous crises — particularly those that involve a shooter. We anticipate that some frames will persist as more evidence is produced, such as criticism of the Secret Service or blame of mainstream media for inciting violence against a candidate. Others fade more quickly as new evidence emerges, for example about the identity of the shooter. Given the aesthetic affordances of social media platforms, cultural symbols emerging in this moment, ranging from Trump asking for his shoes to putting his fist up in the air, become crystalized in different subcultures and can be used for commercial and political gain by crisis entrepreneurs and political campaigns.  We observe generative AI taking an increasingly prominent role in speeding up the creation and proliferation of such cultural symbols.

Example 1: Speculation of Shooter’s Identity and Motives

The first questions to emerge in collective sensemaking during shooting crises are likely to be “Who did this and why?”_[1] Our analysis reveals that, in the wake of the assassination attempt of Donald Trump, rumors about the shooter’s identity began to spread within 5 minutes of the shooting on X (formerly Twitter), and soon echoed across other platforms. On Telegram, the shooter identity rumors took shape across two distinct stages (Figure 1). 

Regarding the Who? In the first stage, prior to the shooter’s identity being officially released, public Telegram channels prematurely pointed the finger at an alleged “Antifa” member. Following this initial false identification, an individual on X posted a video falsely claiming to be the shooter, leading to more confusion. In subsequent fact checks, the self-identified shooter was identified as an internet troll (with an antisemitic handle) who later claimed he posted the video as a joke.

Regarding the Why? Once the actual shooter’s identity was officially confirmed and these initial false claims and speculations proved incorrect, rumoring entered a second stage focused entirely on theorizing about the shooter’s motives _[2]. Telegram channels featured accusations and insinuations about the shooter’s potential connections to the Democratic party, presenting reports that he previously donated to a progressive fundraising group through the ActBlue donation platform. Elsewhere, media outlets noted that public records suggested he was a registered Republican, but this piece of evidence was not salient on Telegram, where U.S. audiences are highly skewed towards conservative politics.

Across both Stage 1 and Stage 2, we noted how the production of evidence on Telegram reflected the use — and promotion — of a strategic political frame that sought to paint the perpetrator as motivated by a left-wing agenda and/or anti-Trump rhetoric. That frame stayed consistent even as the available facts hinting at the shooter’s identity shifted.

Figure 1: An anonymized cumulative graph showing the number of posts, over time (UTC), about the shooter’s identity. Circles represent public Telegram channels, sized by the number of channel subscribers. Original posts are marked in pink, while forwarded posts are navy. This graph was generated using a continuously updated sample of public Telegram channels posting about U.S. elections. The sample was constructed using a snowballing method by following Telegram “forwards.” The initial seed list from which we snowballed consisted of 1) the largest U.S.-focused channels listed in the online Telegram catalog TGStat and 2) additional channels identified through manual monitoring and investigation as focused on U.S. elections.

As right-leaning audiences on Telegram focused on framing the shooter as a left-wing activist, left-leaning audiences elsewhere (e.g., Bluesky and X) approached the task of understanding the shooter’s motives from the opposite point of view.  Conflicting rumors circulated (and continue to circulate) around the shooters’ past political activities and actions, including that he was a registered Republican, had donated to a progressive political cause, and that he appeared in a Blackrock commercial_[3].

Framing competitions that attempt to lay “blame” on one political party or another for motivating a shooter are not uncommon. They were also seen after the Tree of Life Synagogue shooting in Pittsburgh, as rumors circulated about the assailant’s party membership, as well as his anti-semitismic and xenophobic posts on the social media platform, Gab. It is important to note that even in assassinations of prominent political figures, the primary motive is not always directly political, such as in the assassination of former Japanese prime minister Shinzo Abe in 2022. Investigating the factors contributing to episodes of political violence, including the assailant’s motives, takes time. This ambiguity leaves space for different groups to spread identity-based frames for interpreting evidence about the shooter and his motivations. We have seen examples of false identity-based frames in other recent U.S. shootings, such as false claims that both the 2022 Uvalde, Texas and the 2015 Colorado Springs, Colorado shootings were perpetrated by transgender people. In the coming days, we expect more online rumoring and speculation about these motives. 

Example 2: A Spectrum of Political Frames

In the many cases of online collective sensemaking that we’ve studied, the application of politically-coded frames is a salient part of the process. Across the political spectrum, commentators often use the same videos, images, eyewitness accounts, and “expert” analyses to support disparate rumors and theories aligned with their political preferences. 

In the first days after the assassination attempt, we observed three core politically coded frames emerge. On the anti-Trump extreme, there was an early frame that the event was staged for Trump’s political gain, though this has since slowed as more evidence emerged (visible in Figure 2). Another general frame emerged criticizing and questioning the Secret Service, with some partisan influencers applying an ongoing critique of Diversity, Equity, and Inclusion (DEI) initiatives to explain the Secret Service’s failure — a frame we’ve seen in other recent crisis events. On the pro-Trump extreme, a frame emerged that the assassination attempt had been an “inside job.” Both extremes reflect more conspiratorial framing devices and instincts. 

“Staged” Assassination Attempt

In the early aftermath, a frame took hold on TikTok, X, and Bluesky that the assassination attempt was staged for Trump’s political benefit. Commentators insinuated that the event was set up to portray him as a hero with photographers strategically positioned to capture the instantly iconic photo of him with blood on his ear and his fist in the air. Others claimed that the crowds not panicking enough was “proof” that they were crisis actors. Some pointed to Trump’s history with World Wrestling Entertainment (WWE) and reality TV — suggesting he had “bladed” or cut himself like pro-wrestlers do for dramatic effect — as indication that this was a professionally staged event. Mimicking “expert” testimony, some professionals in the entertainment industry compared the blood on Trump’s ear to the kind found in packages on studio lots. Spanish-language videos drew on experiences of “fake crimes” in other Latin American countries to bolster the “staged” claim.  

 

Figure 2: A temporal plot showing the number of posts on X that discussed the attempted assassination using the word “staged”. The x axis is Time (EDT) and the y axis reflects the number of estimated tweets related to a set of keywords about the assassination attempted, narrowed to posts containing “staged”. The data in this graph was visualized by using Brandwatch Consumer Research.

Theories derived from this frame peaked in the first two hours after the shooting, but died down shortly thereafter, as reputable news outlets had reported this as a legitimate assassination attempt and the shooter had been identified. Interestingly, though the “staging” frame was prominent this time within communities on the political left, it resonates with conspiratorial frames that are repeatedly used by the right-wing audiences to process crisis events, such as “false flag” and “crisis actors” accusations, where an event is produced in order to facilitate a political or economic victory. 

Secret Service Failures

On social and mainstream media, many across the political spectrum, including President Biden, questioned why the Secret Service had failed in protecting a presidential candidate. Social media commentary was particularly focused on the failure to secure the rooftop where the shooter had carried out his deed. News brokers shared videos of the crowd urging law enforcement to address the armed man on the roof during the rally. Reflecting a more politically-coded frame, some compared videos of the Secret Service’s swift reaction to the 1981 assassination attempt on President Ronald Reagan, suggesting the Secret Service reaction to shots at Trump was much slower.

Many pro-Trump influencers suggested Trump had been denied additional protection — sometimes directly or indirectly blaming it on political motivations. Some spread the rumor that the Secret Service had denied Trump’s additional security requests, which the GOP had been investigating — a claim the Secret Service has denied. This narrative was further fueled by recent proposed legislation calling for the removal of Trump’s Secret Service protection if he were convicted.

Some pro-Trump influencers also blamed the Secret Service failure on DEI initiatives supposedly positioning unqualified minority hires in leadership roles — a frame we’ve seen employed before, spuriously blaming DEI for crises such as the Baltimore bridge collision and Boeing’s subpar manufacturing. Some juxtaposed images of female Secret Service agents at the Trump rally with older images of male agents. We expect this frame to continue gaining traction online. 

“Inside Job” Claims

Some critiques of the Secret Service took on more conspiratorial framing. In retweeting a popular pro-Trump influencer, Elon Musk speculated that the mistake was either “incompetence” or “deliberate” (a framing that emerged on Telegram as well, see below). Highlighting many of the same critiques and questions of how the shooter could get to an unsecured rooftop, other influencers suggested the shooting must have been an “inside job.”  One popular post tried to make sense of how a 20-year-old could outsmart the Secret Service, warnings from the crowd, or the speed of law enforcement’s reaction, only to conclude by comparing the event to the Kennedy assassination and insinuating the failure was potentially intentional. Others spread the rumor that a counter-sniper was told to wait. These speculations align with claims of “false flag” operations in previous crisis events.

While the “staged” assassination attempt, Secret Service failures, and “inside job” rumors were prominent on other platforms, notably X, they remained relatively low level within the sample of data we examined on Telegram — which has often hosted that kind of conspiratorial framing in the past. Much of the conversation in our Telegram data continued to focus on the identity of the shooter and his political affiliation. An important limitation to note here, however, is that our Telegram data is constructed from a snowball sample of U.S.-focused public Telegram channels, and is thus not necessarily representative of the conversation across the entire Telegram ecosystem.

Figure 3: A temporal plot showing the number of posts by public Telegram channels in our sample per hour for four prominent rumors. The ‘“staged” assassination attempt, Secret Service failures, and “Inside job” claims were present among public Telegram channels, but the volume of these rumors remained relatively low-level in our sample of data.

Example 3: Right-Wing Critiques of Mainstream Media 

One of the most salient frames for breaking news and crisis events deployed by the right-wing is criticism of mainstream media. Historian AJ Bauer has described opposition to mainstream media as a defining and uniting characteristic of post-War conservatism. We saw this frame deployed in a variety of critiques including claims that mainstream media outlets had incorrectly reported the event; that the media had been too slow to call the shooting an assassination attempt; or that the media was partly responsible for goading the public into a frenzy against Trump by claiming he is a threat to democracy. The frame of mainstream media opposition is so strong among many Trump supporters that some rally goers were recorded shouting “This is your fault!” at the press box after the shooting. 

As crises unfold, media institutions are subject to some of the same collective sensemaking dynamics as the rest of the public. It can take time for journalists to generate reliable leads and release accurate reporting — and in rapidly evolving contexts like this one, sometimes incorrect or frustratingly vague information is initially reported. This dynamic served as another vector for conservative criticism of mainstream media. In the first reports that something was wrong at Trump’s Pennsylvania rally, headlines stated that Trump had fallen or been escorted away in response to loud noises — drawing ire from online commentators who claimed the media was trying to downplay the severity of the event. Though these vague initial headlines were later updated after credible sources confirmed Trump had been targeted in an assassination attempt, screenshots highlighting the hedgy language in the initial headlines persisted on social media.

Subsequent rumors, including one tweet featuring a video clip of French far-right politician Marine Le Pen, criticized what was perceived as media hesitancy to call the event an assassination attempt.  As these critiques gained prominence, they furthered the longstanding delegitimization of the media. Social media users, including Elon Musk himself, championed X as a quicker, more reliable, and less partisan source of information for breaking news than an outlet like CNN. 

Beyond the criticism of initial reporting, social media users blamed the mainstream media for motivating the attack, suggesting previous negative coverage about Trump had stoked virulent anti-Trump sentiments amongst the American public. They particularly lambasted legacy media personalities as victim blamers, who pointed out Trump and his supporters had warned of political violence as a consequence of Trump’s legal prosecutions. Some social media users shared compilations of historic headlines in which Trump was criticized, meanwhile prominent figures like U.S. Rep. Marjorie Taylor Greene blamed the media for “demonizing” Trump.

As the circumstances around the shooting continue to unfold, screenshots of previous reporting will continue to appear incorrect or outdated. Though continuously refining a developing story is a normal part of the journalistic process, it may continue to serve as a vector for delegitimization of mainstream media. 

These rumors and criticisms reveal some of the challenges faced by professional journalists grappling with both legitimate criticism and politically-driven villainization of their work: they are criticized for both moving too slow and expressing uncertainty when they report quickly, also vilified for making mistakes and having to correct themselves, and criticized (often by actors on both sides of the political spectrum) for being biased.

Example 4: Cultural and Commercial Production During Crises

Oftentimes, the trendiest content does not offer a particular verbal or textual explanation for a crisis, but instead evokes an affective response to a resonant aesthetic produced through trending sounds, images, and pop cultural references, and may spread in a variety of intersecting subcultures (e.g., partisan groups, fandoms, or racial/ethnic identities). The ability to create these aesthetic connections is easier than ever due to the emergence of computational tools like generative AI. The power of imagery, symbols, and sounds to assign ambiguous but powerful emotional resonance in pursuit of political or social mobilization — or commercial gain via commemorative merchandise — must not be overlooked in moments of crisis. 

Humor and Pop Culture to Process Crisis

On contemporary social media, pop cultural references may be used to memorialize, explain or make light of crisis events. We saw memes comparing Trump’s grazing of the bullets to the Matrix films bullet dodges, alluding to Simpsons predictions, or even using Spongebob characters to depict a jealous Abraham Lincoln. Satirical content can also act as frames, such as “interviewing the assailant’s gun” which served the rumor that the event was staged and had crisis actors.  Some content elevates the moment to iconic status, like celebrating what sounds like Trump saying “let me get my shoes” as the Secret Service ushers him away. Reactionary meme production is a hallmark of any cultural event in the internet era as creators try to vie for attention during a trending moment or cope with what has happened.

Fandoms and Subcultures Cement Group Identity Via Cultural Interpretations 

Different ethnic, fandom, and cultural groups often react to the event with inside references. For instance, the Fortnite community created reenactments in the game and posted videos of it. Perhaps most visible even outside of its own community is Black Twitter_[4] and Black TikTok’s reactions ranging from relief at the shooter being white to reenacting the scene to asking Shonda Rhimes to come back and write a Scandal television episode.

In particular, “Many Men (Wish Death)” by 50 Cent — a song about a man who cannot die despite hits on him — trended as folks celebrated Trump’s survival. Many videos on TikTok utilized the track to create videos about Trump, which quickly resulted in a remix of Many Men on several platforms and the acknowledgement of the meme by 50 Cent at a performance.  The song was a way to signal one’s support for Trump and find others who do — especially because users can search for more content that has that viral sound or song. Rumors have been reported that 50 Cent may perform at the Republican National Convention — underscoring how online content can impact offline political organizing.

AI Fueling the Generation of Content and Commerce 

Generative AI may be fueling the faster generation of cultural and commercial content for creators. The mass amount of sharing and creating may contribute to internet trends and a shared “vibe,” or affective sentiment, within an app. Some of the content may appear to be or is considered “AI Spam,” which has gained popularity online and identified in a recognizable style (like the image below of Trump emerging from clouds) as mass-produced content is made to generate engagement and downstream profits. The rapid production of a shared aesthetic language and collective zeitgeist may result in drawing connections to other related frames more quickly — an area where more research could be done because of the newness of generative AI.  

Additionally, many users increasingly leverage AI to quickly output listings for products they will sell with the attention they receive by jumping on the trending event. Capitalizing on crisis events is not new, but what is novel is the near instantaneous ability to commemorate it by quickly selling merchandise (that may not exist yet, or ever) with product images or listings generated by AI. We can anticipate this phenomenon much more during this and future crises. 

Commercial incentives — including merchandise and monetization from mass followings — can also contribute to the propagation of a shared sentiment in the wake of a crisis. This is an essential component to the influencer attention economy.  The ubiquity of commemorative T-shirts or tattoos is so common that users were quickly joking about how long it would take for these shirts to be made and posting when the shirts were out.

Commemorative merchandise contributes to mobilizing and demonstrating support for Trump in this moment. Product listings relating to this event vary from shirts to phone cases, with several products related to this event appearing on in-app shops like TikTok Shop and Facebook Marketplace, as well as the platform Etsy. Along with in-app marketplaces, there are digital ads being run on several platforms like X to advertise a wide range of products including collectible commemorative trading cards, digital currencies with $MAGA surging after these events. Some of these products may be inauthentic listings and the public should be mindful of the incentives of crisis entrepreneurs. 

 

 

The Humanity of Rumors

Rumors are a natural part of human life, especially during moments of crisis when emotions are high and the facts are uncertain and changing. Researchers understand collective sensemaking — which produces rumors — as a natural social process, and one in which we all partake. Under these conditions, we all make mistakes. Getting caught up in conspiracy theorizing during the aftermath of the attempted assassination of a major political figure is a common human response. Collective sensemaking processes are also subject to manipulation. Bad actors and hoaxers can introduce faulty evidence. Political operatives may attempt to bend the frames that people use to select and interpret evidence. Scammers and crisis entrepreneurs can try to exploit heightened emotional states to spread falsehoods, sell products, and gather donations. Our hope in writing this article is not to scold people for sharing rumors, but to help people understand our very human vulnerabilities, especially to intentional manipulation, during times of crisis and social disruption.

Additionally, we aim to shed light on how the sociotechnical structures and dynamics of online information spaces may make us even more vulnerable, under these conditions, to spreading false rumors and to being manipulated by scammers and political point-scorers. Recognizing how our political identities are intentionally exploited — and even just incidentally make us susceptible — to spread false rumors may help us become more resilient to these forces. Slowing down and paying attention to which frames we are using and how they shape which evidence we focus on and how we interpret that evidence may help us in navigating complex digital spaces. 

---

• Danielle Lee Tomson is the research manager at the University of Washington’s Center for an Informed Public.
• Kate Starbird is a CIP co-founder and Professor in the UW Department of Human Centered Design & Engineering.
• Nina Lutz is a CIP graduate research assistant and UW HCDE doctoral student.
• Stephen Prochaska is a CIP graduate research assistant and UW Information School doctoral student. 
• Ashlyn B. Aske is a CIP graduate research assistant and UW School of Law Master of Jurisprudence student.
• Melinda McClure Haughey is a CIP graduate research assistant and UW HCDE doctoral student. 
• Joseph S. Schafer is a CIP graduate research assistant and UW HCDE doctoral student.
• Zarine Kharazian is a CIP graduate research assistant and UW HCDE doctoral student. 
• Adiza Awwal is a CIP graduate research assistant and UW HCDE doctoral student.
• Michael Grass is the CIP’s assistant director for communications.

---

Footnotes

• [1] For example, in the aftermath of the 2013 Boston Marathon bombings, social media users mobilized to search for clues to the identities of the suspects, resulting in multiple false identifications.
• [2] At the time of writing this blog post, no official motive has been reported in the shooting at the rally.
• [3] The financial services company has been subject to right-wing attacks for its support of Environment, Social and Governance (ESG) initiatives.
• [4] A Note on Terminology: Following a common practice in academic scholarship, we refer to the different elements of X (tweets, retweets, likes, follows) by the terms used in the community itself. Likewise, we use terms like “news twitter” and “crisis twitter,” both lower case, as the common names used on X to refer to these communities or discourse subgroups. When referring to the platform itself or associated policies, we use the term “X,” the legal name of the platform.