Check out the Tackling Hate Lab’s new report on anti-First Nations hate online in Australia. Using more than three million posts collected across social media platforms, news websites, forums and blogs, the report provides one of the largest analyses to date of how online hate targeting Aboriginal and Torres Strait Islander peoples changes across major national events. It compares two very different contexts: the 2023 Voice to Parliament referendum campaign and two events in 2025, Anzac Day and March for Australia. Together, the datasets include over 3 million online posts created by more than 370,000 unique accounts, allowing us to examine not only how hate is expressed, but also how it spreads and changes over time.
The report combines large scale data collection with community informed artificial intelligence. We first trained an AI model using almost 2,900 social media posts manually labelled by trained community annotators as hateful or non-hateful. We then used a second AI system to classify different forms of anti-First Nations hate and identify the narratives they contained. This approach makes it possible to analyse millions of posts while ensuring that the classification reflects how First Nations communities themselves understand and experience online hate.
The findings show that anti First Nations hate is highly shaped by context. During the Voice referendum, most hate focused on opposition to the Voice itself, while in 2025 hate shifted towards direct attacks on First Nations identity, including White supremacist and dehumanising narratives. The report also shows that misinformation, political events and media attention can amplify hate, and that First Nations public figures are disproportionately targeted. These findings demonstrate the need for continuous monitoring, stronger responses to harmful misinformation, better protection for targeted communities, and greater accountability from digital platforms.