A Year of Hate: Antisemitism on Telegram Post-10/7

Introduction

October 7, 2023, was the deadliest day for Jews since the Holocaust.

Hamas, a violent Palestinian Islamist organization with a history of antisemitism, along with other Palestinian terror groups, carried out a series of attacks against Israel. More than 1,000 Israelis were killed, of whom a majority were civilians. Numerous cases of atrocities by Hamas members included rape, mutilation and the kidnapping of 251 people, some of whom were children and elderly individuals.

Previous ADL Center on Extremism (COE) analyses have shown that acts of violence against Jews and Israelis can lead to upticks in antisemitic discourse. To understand the longer-term impact of the 10/7 attacks, COE analyzed online violent antisemitism and anti-Israel content on Telegram, a large private messaging service and social media platform that attracts extremists because of its privacy protections and lack of restrictions on hate speech.

COE conducted a comprehensive analysis of English-language posts on the social media platform Telegram to assess the influence of these attacks on the level of violent online antisemitism and anti-Israel content in the subsequent weeks and months. These three terms -- antisemitism, violent online antisemitism and extreme antisemitism -- are used interchangeably in this report to describe direct threats and calls for violence, as well as dehumanizing language towards Jews and Jewish identity. COE continuously monitors several thousand Telegram channels, and for this report, researchers used a combination of qualitative, quantitative and machine learning techniques to analyze the volume, loci and reach of violent antisemitic activity on Telegram over the past year.

The results of the analysis covering April 2023-April 2024 answer three questions:

1. How has the volume of extreme antisemitism on Telegram evolved following the 10/7 attacks?
2. Where are extreme antisemitic messages taking place—that is, on which channels and in what concentrations? How has this changed over time?
3. How has engagement with extreme antisemitic content changed since the attacks? How many users are actively responding to—or participating in conversations containing—violent antisemitic messages?

Using a multi-method approach that includes machine learning models and qualitative investigations, we provide evidence for an initial massive spike in violent antisemitic activity. However, while this spike in antisemitism subsided relatively quickly, antisemitic activity remains persistently high relative to before October 7, suggesting a “new normal.”

Key Findings

1. Immediate Spike in Antisemitic Activity: The day of the attacks saw a 433.76% increase in violent antisemitic posts on Telegram, jumping from an average of 238.12 posts per day to 1271 posts. This surge, though diminished over time, has resulted in a significantly higher average daily volume of antisemitic posts (321.22) compared to pre-attack levels.

2. Sustained Elevated Levels: Post-attack, the level of violent antisemitism has not returned to pre-October 7 levels, indicating a new, persistently higher baseline.

3. Concentration in Key Channels: The increase in antisemitic activity was notably concentrated in a subset of Telegram channels known for extremism. Channels like GHOSTCHAT and GDL Chat 2.0 consistently host the most violent antisemitic content.

4. Increased User Engagement: The number of unique users actively forwarding violent antisemitic messages rose weekly from 12.18 to 20.96, with the volume increasing from 204.03 to 346.4 per week.

5. Causal Relationship: Through a difference-in-differences analysis, using content from the sovereign citizen movement as a control group, the study provides strong evidence that the October 7 attacks directly caused the rise in antisemitic activity. (Difference-in-differences analysis compares changes in an outcome variable between a treatment group and a control group before and after an intervention).

6. Qualitative Support: Numerous posts explicitly referenced the October 7 attacks, supporting the conclusion that these events were a significant driver of the observed increase in violent antisemitism.

7. Emboldened Existing Users: Active user growth following the immediate aftermath of October 7 was modest and confined to a small number of channels, suggesting that increased antisemitic content is being driven predominantly by emboldened existing users, rather than a large influx of new ones.

The findings indicate a concerning and sustained rise in violent antisemitism online following the October 7 attacks, suggesting a new normal in the baseline level of such activity on platforms like Telegram.

Telegram’s failure to address extremism and violent antisemitism on its platform is concerning. The data suggest that users who hold antisemitic views are feeling emboldened to express them more frequently and forthrightly.

Data and Results

Telegram is a digital platform and messaging service with hundreds of millions of monthly active users, social media-resembling channels and groups, and forums containing sometimes tens of thousands of users who rely on Telegram’s anonymity and privacy protections. These features have helped marginalized individuals—such as journalists in authoritarian countries—avoid government repression. But they are also exploited by those who want to harass, threaten and spread extreme ideologies. Because Telegram does not typically engage in content moderation, the level of hateful and extremist content posted on the platform can be high.

Following the arrest of Telegram CEO Pavel Durov in France on August 25, 2024, on allegations that criminal activity was allowed on the platform, there was initial optimism that the platform might implement moderation guardrails. As of September 2024, no significant public facing changes have been made in addressing online hate and antisemitism on the platform. COE routinely monitors around 2,000 channels and chats known to play host to extreme ideologies.

This analysis draws on 12.2 million total messages and 169,080 unique Telegram users. Table 1 describes our data. There were, on average, 278.40 extreme antisemitic messages every day, for a total of 101,895 between April 2023 and April 2024, accounting for 0.83% of all messages we assessed during that period. This fraction increases to 1.03% if we restrict attention to text messages, of which there were around 9.8 million. Of the antisemitic posts, 23,523—or 23 percent—contained a media attachment.

Table 2 shows the current sample of 793 channels and chats—those that posted at least one violent antisemitic or anti-Israel message between April 2023 and 2024. While the number of posts differed wildly across channels, the channel with the most antisemitic posts, “GHOSTCHAT,” contained 30,268, or nearly one-third of all such messages.

Table 3 focuses on user behavior with 7,769 users posting at least one antisemitic message during the study period. The distribution of these posts is notably skewed: among users who shared antisemitic content, the average number of such messages was 10.78. However, the high standard deviation of 66.71 indicates significant variation in posting frequency among these users.

Findings

This report uses several different methodologies, including qualitative analysis, multiple statistical tests and machine learning to detect violent and threatening antisemitic speech. Findings fall into three broad categories of metrics: volume of extreme antisemitic posts, top channels based on those experiencing growth in antisemitic content and engagement by message forwarding.

Post volume

In the period before October 7, 2023 (from April 1–October 6), on average 238.12 violent antisemitic posts were posted each day. On October 7, that number shot up to 1271 posts, a 433.76% increase. Subsequently, the daily average decreased in fits and starts, but remained well above the pre-10/7 level, at 321.22 average daily messages, a 34.8% sustained and statistically significant increase concerning the six months preceding 10/7. Figure 1 shows the number of antisemitic posts as a percentage of all monitored content.

We also provide preliminary evidence of the causal impact of the events of October 7th through a difference-in-differences (DiD) analysis. This method allows us to estimate the effect of October 7th on violent antisemitic posts by comparing changes in posting behavior before and after the event while using content associated with sovereign citizen ideology as a control group.

The DiD approach works by comparing the difference in daily average posts for violent antisemitism before and after October 7th with the difference observed in the control group over the same period. This helps isolate the specific impact of October 7th from other factors that might have influenced posting behavior.

The outcome of this analysis, as shown in Table 4, reveals that October 7th led to an average increase of 83.34 violent antisemitic posts per day, derived by subtracting the change observed in the control group (a slight decrease of 0.24 posts per day) from the change in the group of interest (an increase of 83.1 posts per day).

Reach/engagement

The impact of antisemitic content extends beyond frequency of posts and depends on user engagement, such as how often these posts are forwarded.

On the channels and chats monitored by COE, there were an average of 12.18 unique weekly forwarders (individual users) of extreme antisemitic posts prior to October 7th. After October 7th, this number increased significantly to 20.96 (or 19.71 when excluding the week immediately following October 7th).

Similarly, the average number of violent antisemitic and anti-Israel messages forwarded weekly increased from 204.03 to 346.4 (320.63 when excluding October 7th and the following week). Figure 2 illustrates these differences with a bar chart.

Top channels/chats

These are the fastest-growing chats and channels for violent antisemitism.

Some of these communities are among the usual suspects for antisemitic activity and figure prominently in the list of channels where antisemitism shot up following Iran’s April 2024 attack against Israel. Some of the top-growing antisemitic chats include:

• Professional cyberbullying Russians [here name changed to “Frank”] is an online Telegram chat originally created after the start of the Russian-Ukrainian war focused on harassing Russian soldiers and supporters. The channel allows and encourages far-right, antisemitic and extremist views.
• Amalek International is a Telegram channel that aggregates antisemitic and white supremacist content from a range of sources.
• EMJ Live Chat airs live chats with Holocaust denier E. Michael Jones. It is run by Culture Wars Media, a magazine and website run by Jones.

Figure 3 compares user and post growth for a selection of the top Telegram chats and channels for antisemitic content. They have all experienced striking upturns in antisemitic posts, while user growth has been less consistent across channels.

Conclusion

This report points to a troubling proliferation of violent antisemitic content on Telegram following the October 7, 2023, attacks on Israel and should be a wake-up call to Telegram leadership.  This research demonstrates what happens when a tech company does not take even the most basic steps to address the proliferation of hate and extremism on its platform: the creation and growth of entire channels, comprising thousands of users, whose purpose is to disseminate threats and hate against Jews and other marginalized groups. At present very little of the heinous content analyzed as part of this research is actually violative of Telegram’s stated policies. Telegram has no policy prohibiting hate. It has no policy prohibiting extremism. Users in private spaces on Telegram who want to report that a channel has devolved into grotesque displays of violent antisemitism, racism, and misogyny have minimal recourse at present.

In a recent interview with Tucker Carlson, Telegram CEO Pavel Durov stated that he wanted Telegram to be a “neutral” platform. The decisions made by Telegram to not invest in content moderation and disregard the safety of marginalized groups are not “neutral” decisions- they are ones that embolden and amplify the worst actors in society- antisemites, hate mongers and extremists. Fortunately, the data do not show a large influx of new antisemitic users following October 7 but suggest that a vocal minority of users are feeling emboldened since the attacks.

These findings should also be of note to lawmakers interested in reforming policy on social media to address antisemitism and other kinds of hate speech.

Telegram’s lack of robust content moderation and idiosyncratic channel structure tends to foster antisemitic content. This is not unique, but Telegram does illustrate the importance of platform interface and moderation policy in controlling the publicly visible antisemitism.

This report provides important insights into the consequences of October 7. It provides evidence of a higher “new normal” —a level of antisemitism greater on average than before the 10/7 attacks which calls into question the conventional wisdom that terrorist attacks have transitory effects on outcomes like hate speech. Antisemitism on Telegram rose following the deadliest attack against Jews and Israel in decades.

These results should be deeply concerning to citizens who care about the spread of violent antisemitism on online messaging platforms like Telegram. Recent history shows that hate that begins online does not end online.