Australia has spent much of the past two years responding to anti-Jewish and anti-Muslim hate as separate problems. But our latest research suggests they have something important in common.
Both changed fundamentally after the Hamas attacks of October 7 2023 and the subsequent war in Gaza. Both became more persistent online. Both became much more closely connected to events happening in the real world.
That matters because Australia is now debating how best to respond. The Royal Commission into Antisemitism and Social Cohesion is underway, and the federal government has appointed Special Envoys to tackle antisemitism and Islamophobia.
Today, the Tackling Hate Lab, a research group bringing together experts in social science, psychology, data science, engineering and economics, released two reports examining these changes.
We analysed more than two million Australian social media posts, most of them from X (formerly Twitter), together with hundreds of verified offline incidents of vandalism, harassment and physical violence targeting Jewish and Muslim Australians between 2021 and 2026.
The findings suggest Australia is dealing with something more than occasional spikes in hate. They point to a lasting shift in how hate spreads and how it interacts with events in the real world.
Our research
The biggest surprise was not that hate increased. It was that the increase persisted.
People often disagree about how anti-Jewish and anti-Muslim hate should be defined. Some definitions are broader than others.
To make sure our findings did not depend on one particular definition, we measured both forms of hate in several different ways. Our data shows that every approach told the same story.
For example, across the two reports, we used Google’s Perspective API to identify posts containing identity attacks. This means language that insults, demeans or dehumanises people because of who they are.
We then used the large language model Qwen to identify who was being targeted, such as Jews, Israelis, Zionists, Muslims and Palestinians.
From shock to system
Identity attacks targeting Jews increased from an average of 0.3 posts per day the year before October 7 2023 to an average of 16.8 posts per day the year after. It remained at a higher baseline until March 2026 (when we stopped collecting data).
Identity attacks targeting Israel increased from 1.3 to 78.9 posts per day over the same period.
Identity attacks targeting Muslims followed a similar pattern, increasing from an average of 2.8 posts per day before October 7 to 42 posts per day in the year afterwards.
For anti-Muslim hate, however, the largest surge came after the Bondi terrorist attack, when identity attacks targeting Muslims reached an average of 1,323.3 posts per day in the week following the attacks, before stabilising at around 224.4 posts per day in the subsequent month.
The same trend appeared regardless of how we measured hate. October 7 shifted the baseline, and Bondi marked a huge increase in anti-Muslim hate and a smaller spike in anti-Jewish hate.
It appeared when we looked at identity attacks, toxic language and several different target groups (including Muslims, Palestinians, Jews, Israel, Zionists).
It also appeared when we used an artificial intelligence (AI) model we developed specifically to reflect how members of the Australian Jewish and Muslim communities perceive anti-Jewish and anti-Muslim hate online. The trends were the same, but the size of the increase varied depending on the method.
That is expected because each approach measures something slightly different. What matters is that every method showed the same long term shift.
This is important because public debate in Australia often focuses on definitions, but our findings suggest that, whatever definition is used, the overall trend remains remarkably consistent.
One important caveat is that our anti-Jewish and anti-Muslim datasets were built differently because they captured different types of online discussions. This means we can compare trends over time within each community, but we cannot directly compare the overall amount of anti-Jewish and anti-Muslim hate or conclude that one was more common than the other.
The vicious cycle
In our reports, we also wanted to understand the relationship between online hate and hate incidents happening in everyday life in Australia.
One of our clearest findings was that they were increasingly connected. Before October 7, online discussions and offline incidents were often separate.
Afterwards, attacks targeting Jewish or Muslim Australians were much more likely to trigger waves of hostility online.
Bianca De Marchi/AAP
Incidents targeting Jewish and Muslim people and properties such as vandalism, harassment and physical attacks increasingly generated spikes in hateful online discussion. For example, in our dataset, every verified offline incident of anti-Muslim hate generated an average of about 12 additional hateful posts in online discussions.
This matters because many of these incidents received widespread news coverage. While most reporting aimed to inform the public, when the stories were shared online, it also created opportunities for a minority of people to spread abuse, conspiracy theories and hostility through comments and social media discussions.
As a result, online hate and offline hate can no longer be treated as separate problems. They have become part of the same cycle, with events in the real world increasingly driving reactions online.
What should happen next?
The Royal Commission into Antisemitism and Social Cohesion and the appointment of Special Envoys are important steps. But Australia also needs better long-term evidence.
At present, governments often collect detailed evidence only after major crises. That makes it difficult to identify emerging problems before they escalate.
We believe Australia needs a permanent Observatory of Hate that monitors both online and offline hate targeting all communities, not only Jewish and Muslim Australians. This should include hate targeting First Nations people, migrants, women, people with disabilities and people of diverse genders and sexualities, among others.
Read more:
The antisemitism debate is already a political minefield. The royal commission must rise above it
A permanent observatory would allow governments, researchers and community organisations to understand how hate changes over time. It could help identify emerging threats, detect coordinated networks spreading hate, evaluate whether policies are working, and improve accountability for technology companies such as X, Meta and TikTok.
Most importantly, it would help Australia move from reacting to crises towards preventing them.
The biggest lesson from our research is not simply that anti-Jewish hate and anti-Muslim hate increased after October 7. It is that both appear to have entered a new phase.
They became more persistent. They became more closely connected to events in everyday life. And they became capable of escalating very quickly after major incidents.
Recognising that shift is the first step towards designing policies that prevent harm rather than simply responding after it occurs.
