A recent report published by the Center for Strategic and International Studies (CSIS), “Left-Wing Terrorism and Political Violence in the United States,” risks feeding false narratives about political violence and polarization. The report’s authors, Daniel Byman and Riley McCabe, make a sweeping claim: 2025 is on “pace to be the left’s most violent year in more than three decades” and left-wing terrorism is “on track…to reach historically high levels.” The evidence used to sound this alarm consists of just five plots and attacks that occurred over a nearly seven-month period this year. According to the data presented in the report, these events represent a 400 percent increase in far-left plots and attacks from last year.
But the CSIS study suffers from fatal analytic flaws. For starters, like shark attacks, the number of events attributed to left-wing terrorism this year is so low in absolute terms that it simply does not justify inducing panic with eye-popping headlines.
Indeed, these five events are doing a lot of heavy lifting in Byman and McCabe’s analysis. They are given an unwarranted level of causal and predictive power. For instance, while the report clearly specifies the authors are analyzing attacks “so far” in 2025, elements of the data presentation leave the piece open to interpretation, implying these events are a forecast of what is to come. The opening paragraphs alert readers that “2025 marks the first time in more than 30 years that left-wing attacks outnumber those from the far right.” The authors compare the number of left-wing events to both right-wing and jihadist attacks and plots, arguing that the latter two types of threats have declined. This pronouncement is, at best, premature, given that a quarter of the year remains.
It is also misleading. Consider that according to the authors’ own data, 13 victims died as a result of left-wing attacks from 2016 through the first six-plus months of 2025. During that same time period, 82 victims perished in jihadist attacks, while 112 people were killed in right-wing attacks. In fact, 14 people were killed in an ISIS-inspired car ramming attack on Bourbon Street in New Orleans on New Year’s Day 2025 alone – more than all those who perished in left-wing attacks inside the United States in the last nine-plus years.
In addition, in the week following the report’s release there were at least two incidents that, while still under investigation, may meet the authors’ definition of right-wing political violence. The first was an anti-LGBTQ plot to attack a Texas pride parade. The second was an attack on a Michigan Mormon congregation by a man who, by all appearances, was a conservative who supported President Donald Trump, though his motives were not immediately clear. The second attack left four people dead and eight injured, a casualty count that is greater than all five of the events from this year that are attributed to left-wing terrorism in the CSIS report. Additional examples of attacks and plots this year that could be attributed to far-right actors are cited below. When just a few events could virtually eliminate a supposedly strong statistical finding, readers should be skeptical of any predictive claims made from the data.
What Longer Term Trends Tell Us
There are good reasons to be worried about a possible increase in far-left violence, especially as the Trump administration’s actions risk provoking further backlash. However, forecasts must be informed by recent and historical data. On this latter point, there is really no debate: The data have consistently shown a greater threat of political violence from far-right actors.
Indeed, according to the authors’ own data, far-right extremism accounted for an average of approximately 20 plots and attacks per year over the last decade, while far-left extremists were responsible for just four incidents per year during the same period. Even if attacks were to continue at their current pace, far-left extremism in 2025 will not come remotely close to the scale of violence that Americans have become accustomed to experiencing from the far right.
Small differences in incident rates can be quickly wiped out by a single mobilizing issue for a group, movement, or ideology. Even a short-term spike in far-right extremism would eliminate the small gap between far-left and far-right attacks that exists in the authors’ data. More importantly, even if far-left incidents outpace far-right ones this year, the low overall, absolute numbers alongside the far right’s history of violence indicate that it would be foolish to reallocate resources to focus exclusively on far-left threats. While the authors do not advocate for resource allocation exclusively to potential threats from the far left, it is a conclusion that must be more strongly guarded against in a political environment in which the entire American left is being cast as a clear and present danger.
Are Far-Right Attacks Really Declining?
Another fundamental problem with the report is the authors’ failure to clearly explain why some incidents were included and others excluded from their analysis. While the report and its accompanying codebook provide definitions of far-right and far-left terrorism, the authors do not detail how they applied their definitions to real-world violence to make inclusion decisions.
For example, the authors include an allegedly far-left terrorist plot from January in which a woman armed with a knife and two Molotov cocktails told U.S. Capitol police of her desire to attack several administration officials, including Defense Secretary Pete Hegseth, Treasury Secretary Scott Bessent, and House Speaker Mike Johnson. While this was not a particularly sophisticated terrorist scheme, a defensible argument can be made for counting it as a disrupted far-left plot.
But why were seemingly similar events perpetrated by allegedly far-right actors excluded from the authors’ analysis? For instance, Byman and McCabe claim “there was only one right-wing terrorist incident in the United States—the killing of Minnesota state legislator Melissa Hortman and her husband in June.” They apparently exclude from their report a case from March in which a 17-year-old was arrested for allegedly killing his mother and stepfather as part of a larger plot to assassinate President Donald Trump. Although his ultimate target was a Republican president, the teenager was not motivated by far-left ideology. Rather, he was active in neo-Nazi accelerationist communities online that fit squarely into most researchers’ understanding of far-right terrorism.
There were numerous other plots in the first six months of the year that could plausibly be attributed to right-wing actors, but the authors apparently determined did not meet their operational definitions of far-right terrorism. These incidents include: a man allegedly stalking the mayor of Salt Lake City in response to her LGBTQ+ pride flag policies, as well as plots to attack an Islamic center in Arizona, a shopping mall in Washington, and synagogues in Massachusetts. There were also half a dozen school-based attacks and plots that were motivated, at least in part, by neo-Nazi ideology or past far-right attacks that did not find their way into the authors’ data. (The authors “generally” exclude school shootings because they claim the “[a]ttackers…are typically motivated by a mix of personal grievances.” While personal grievances often do play a role in attacks at schools, so do ideological motivations, as evidenced by the far-right beliefs reportedly held by some of the school shooters this year.)
The point here is not to argue that far-right extremists have plotted more attacks this year than those on the far left, although that could be true. Rather, without clearly articulated inclusion rules, the data are untrustworthy. Problematic data are a concern at any time, but particularly when Americans sense that they are witnessing a rapid growth in political violence and are desperate for evidence to help them understand what they are experiencing.
A False Equivalence Between Far-right and Far-left Violence
The potential threat that far-left extremists pose is also artificially inflated in the report by how much substantive significance the authors attributed to the five events from this year. The authors note that historically left actors have been “disorganized” with “limited skill,” as well as limited “effectiveness,” and that their attacks are of “limit[ed] scale and sophistication.” They also correctly note that “the typical target selection, scope and weapon selection of left-wing attackers reflect an intent to signal opposition or cause disruption rather than inflict mass casualties.”
However, paradoxically, the authors then go on to suggest that far-left extremism is a significant threat to public safety because the “recent increase [in plots and attacks] is likely to translate into realized violence.” If left-wing extremists typically do not inflict significant harm, why would recent events portend a future of highly lethal far-left violence? The five incidents detailed in the report do not provide compelling evidence that the far left is becoming increasingly organized and dangerous. Indeed, with the possible exception of a July 4 attack on an ICE detention facility in Texas, the incidents were low in tactical sophistication, were not carried out by organized groups, and were non-lethal. The report nonetheless uses these incidents to draw a moral and strategic equivalence between left-wing and right-wing violence, essentially framing the left as a rising concern equivalent to right-wing extremism that has persistently dominated the threat landscape.
The Complexity and Responsibility of Analyzing Political Violence
It is possible that politically motivated plots and attacks by far-left extremists are on the rise. But objective, robust data are needed to make that determination. Fortunately, there are other data that researchers can analyze to understand the current threat environment, and those sources suggest that it would be a mistake to pin recent increases in plots and attacks on a single group, movement, or ideology.
The Terrorism and Targeted Violence (T2V) in the United States dataset, for example, identified 154 terrorist plots and attacks that occurred in the first six months of 2025.* These incidents represent an 85 percent increase in terrorism when compared to the same period in 2024. The data also show a corresponding 343 percent increase in deaths and a 789 percent increase in injuries from the terrorist attacks that took place during that timeframe.
However, the T2V data, which are made available online for researchers to use and validate, reveal tremendous diversity in the motivations, targets, and ideological leanings of those who committed acts of terrorism this year. For example, there were at least 20 plots or attacks that targeted federal immigration enforcement officers or facilities in the first six months of the year. An argument can be made that these instances should be treated as far-left terrorism, though a closer assessment of each assailant’s motivation is needed to make that determination.
Other incidents in the T2V data were more likely animated by far-right motivations. There were 13 premeditated plots and attacks that targeted peaceful demonstrators who were protesting against these immigration enforcement actions and the broader expansion of executive power — acts that could arguably be categorized as driven by far-right extremism. A closer look at the attackers’ motivations, including any ideological beliefs, is necessary to make a determination.
There were also more than 30 plots and attacks with links to antisemitism, which can be driven by beliefs across the left-right ideological spectrum. Some of these were perpetrated by individuals who were upset over Israel’s actions in Gaza, which Byman and McCabe refer to as ethnonationalist terrorism but other commentators attribute to the far-left. Other antisemitic plots were perpetrated by adherents of far-right neo-Nazism. There were plots and attacks targeting both Republican and Democratic lawmakers. There was also violence perpetrated against LGBTQ+ and Muslim communities and a plethora of property crimes targeting Tesla owners.
Rather than pointing to one ideology as the cause of what feels like an increasingly dangerous environment, T2V’s data point to something worse: growing civil unrest from across the political spectrum that is the result of a vitriolic and mobilizing information environment, the increasingly zero-sum nature of U.S. politics, and the rapid abandonment of targeted violence research and violence prevention programs by the Trump administration.
Researchers have a profound responsibility to not only conduct research ethically but to also share it responsibly. This means considering word choices, avoiding logical fallacies, and clearly spelling out the implications of their work, especially in a highly polarized environment.
It is unsurprising that the CSIS report’s findings were quickly weaponized. For instance, the White House Deputy Press Secretary tweeted an image from an Axios summary of the report with a corresponding caption that blamed Democrats for a “30-year high” in leftwing terrorism.
The post received mocking comments from commentators noting that it ignored how right-wing activity still dominated the last three decades of political violence. (The grey bars in the chart above show attacks attributed to right-wing actors, while the yellow bar shows those attributed to left-wing actors.) Still, many Americans could easily be misled by the headlines generated by the report, especially if they do not evaluate its methodology and findings. The way the CSIS report was used highlights an essential responsibility for researchers in the current politicized era: maintaining methodological rigor and accurate interpretation of data to prevent the weaponization of findings for ideological and political ends.
*One of the authors of this piece, Michael Jensen, was formerly part of the team that produces the T2V dataset.
