This research brief by the Tackling Hate Lab examines how online narratives evolved on Telegram in the first 72 hours following the Bondi attack. The study is based exclusively on publicly accessible English language posts mentioning Arsen Ostrovsky and related keywords. While not a complete account of Telegram activity, the dataset provides a high resolution reconstruction of narrative development in real time.
Using first mention timestamp analysis, hourly frequency tracking, word frequency modelling, and engagement metrics, we identified a layered progression. Discussion began with factual reporting of injury. Within minutes, identity contextualisation appeared. Attention then shifted to the image taken at the scene. Insinuation and image based doubt emerged before direct accusation. Explicit structural conspiracy language such as “false flag” appeared substantially later, followed by circulation of AI generated imagery reframing the event as staged.
The findings show that conspiracy narratives rarely emerge fully formed. They develop through incremental reframing, often driven by posts that blend factual information with insinuation. For policymakers, this highlights the importance of early detection at the identity and image scrutiny stages. For journalists and the public, it underscores how rapidly ordinary reporting can be reinterpreted through online amplification dynamics.
This research brief provided key evidence for this ABC Background Briefing investigation.