Bondi attack came after huge increase in online antisemitism: research

At least 16 people – including a ten-year-old child – are dead after two men opened fire on a crowd of people celebrating the Jewish holiday of Hanukkah on Sunday in a public park at Sydney’s Bondi Beach. Many more are injured.

I am horrified. But as a researcher who studies hate and extremist violence, I am sadly not surprised.

The Jewish community has been a top target for terrorist ideologies and groups for a long time. Many people working in this field have been expecting a serious attack on Australian soil.

Much remains unclear about the Bondi terrorist attacks and it’s too early to speculate about these gunmen specifically. The investigation is ongoing.

But what about antisemitic sentiment more broadly?

Our research – which is in the early stages and yet to be peer reviewed – has recorded a significant and worrying increase in antisemitic sentiment after October 7.

Our research

We have been training AI models to track online sentiment in social media targeting Australian communities, including Jewish people.

That means working with humans – including extremism experts and people in the Jewish community – to label content. This is to teach our model if the content it is encountering is hateful or not.

Based on definitions adopted by the Jewish community, we distinguished between two main types of antisemitism: “old” antisemitism and “new” antisemitism.

“Old” antisemitism targets Jews as Jews. It draws on entrenched myths and stereotypes that portray them as alien, dangerous, or morally corrupt.

“New” antisemitism shifts the focus from individual Jews to the state of Israel. It blames Jews collectively for Israel’s actions.

Many in the Jewish community see this as a modern continuation of historical antisemitism. Critics (both within and outside the Jewish community) contend it risks conflating legitimate opposition to Israeli policies with antisemitism.

Central to this debate is whether anti-Israel sentiment represents a continuation of age-old prejudices or a political response to the Israeli-Palestinian conflict.

In our research, we tracked both “old” and “new” antisemitism.

A sharp increase

We found that both increased sharply after October 7.

For example, we studied posts on X (formerly Twitter) geolocated in Australia before and after October 7. We wanted to understand the size of the rise in antisemitism.

We found that “old” antisemitism rose from an average of 34 tweets a month in the year before October 7 to 2,021 in the following year.

“New” antisemitism increased even more, rising from an average of 505 a month in the year before October 7 to 21,724 in the year after.

Some examples of “old” antisemitism are explicit, such as calls to “get rid of all Jews” or “kill all Jews”.

Others are more indirect, including minimising or denying the Holocaust. Examples include posts claiming that “if the Holocaust of 6 million Jews were true, Israel could not exist today” or that the Nazis had only a minimal impact on the Jewish population.

Other forms of hate rely on conspiracy theories, such as claims that “Jews are paying to destroy Australia”.

However, the vast majority of the content our models identified as antisemitic fell into the category of “new” antisemitism. This included content that blamed the Jewish community for events in Israel, such as calling all Australian Jews “baby killers” or “Zionazi fu–wits”, regardless of their personal political views and opinions about the Israeli government and its actions.

(All examples here are drawn from real content, but the wording has been slightly modified to anonymise them and prevent identification of the original authors).

In other words, we have seen an overall escalation of hostilities against Jews online.

More extreme and explicit calls for violence rarely appear on mainstream platforms. They tend to circulate on fringe social media, such as Telegram.

On X, we have seen a collision of mainstream discourse and fringe discourse, due to the lack of moderation.

But antisemitism doesn’t always involve slurs, meaning it can also happen in mainstream platforms. Especially after the election of Trump and the relaxation in moderation practices of Meta, we have also seen it on Instagram. This includes Instagram posts published after the Bondi attack.

Could more have been done?

Certainly the Jewish community, I am sure, will feel not enough was done.

Jillian Segal, Australia’s first government-appointed special envoy for combating antisemitism, released her plan for addressing the issue back in July.

As I wrote at the time, the recommendations fell into three main categories:

1. preventing violence and crime, including improved coordination between agencies, and new policies aimed at stopping dangerous individuals from entering Australia

2. strengthening protections against hate speech, by regulating all forms of hate, including antisemitism, and increasing oversight of platform policies and algorithms

3. promoting antisemitism-free media, education and cultural spaces, through journalist training, education programs, and conditions on public funding for organisations that promote or fail to address antisemitism.

The government had said it will consider the recommendations. Segal has now said government messaging combating antisemitism has “not been sufficient”.

Some might argue addressing points two and three could have helped prevent the Bondi attack. A common assumption is that a climate of widespread antisemitism can embolden violence.

The reality, however, is that this is hard to establish. People who commit terrorist acts – whether they self radicalise or are recruited by terrorist organisations – do not necessarily respond to changes in broader public sentiment.

That said, there is obvious value in prevention work aimed at reducing hostility and antisemitic attitudes, even while small networks or individuals committed to violent terrorism may still exist.

Preventing terrorist violence of this scale relies primarily on effective law enforcement. This requires adequate resourcing and a clear legislative framework.

Education and broader cultural change matter. In short term, however, they are less likely to be as effective at preventing acts of terrorism as measures such as firearm regulation, monitoring extremist networks, and disrupting plots before they turn into action.