In the distressing hours and days following the terror attack at Bondi beach in Sydney, which left 15 people dead, a parallel crisis unfolded online. Social media platform X, formerly Twitter, became a breeding ground for AI-generated misinformation, with users' feeds flooded by fabricated narratives and digitally altered content.
The Flood of Fabricated Claims
The algorithmically driven "For You" page on X served up a barrage of falsehoods to users seeking factual information. Among the baseless claims were assertions that the tragedy was a staged 'psyop' or false-flag operation, that Israeli Defence Forces (IDF) soldiers were responsible, and that the injured were 'crisis actors'.
In one particularly malicious instance, a deepfaked audio clip of New South Wales Premier Chris Minns was widely shared, featuring a false American-tinged voice making incorrect statements about the attackers. Separately, an AI-generated image manipulated a real photo of a victim, human rights lawyer Arsen Ostrovsky, to suggest he was an actor having fake blood applied.
"I saw these images as I was being prepped to go into surgery today and will not dignify this sick campaign of lies and hate with a response," Ostrovsky later stated on X.
Real-World Harm and International Fallout
The disinformation caused tangible harm to innocent individuals. A Pakistani man living in Australia was wrongly identified online as one of the attackers, a traumatic experience he described as "extremely disturbing". Pakistan's Information Minister, Attaullah Tarar, labelled his country the victim of a coordinated disinformation campaign, alleging it originated in India.
Even the hero of the attack, Syrian-born Ahmed al-Ahmed, was not spared. X's own AI chatbot, Grok, incorrectly told users an IT worker with an English name was the hero who tackled the attacker. This falsehood appears to have originated on a website mimicking a legitimate news outlet, created on the day of the attack. AI-generated images of Ahmed were also used to promote cryptocurrency schemes and fake fundraisers.
Platform Failures and a Broken Fact-Checking Model
The situation marks a stark departure from Twitter's former reputation as a hub for breaking news. While misinformation existed in the past, experts note it is now algorithmically amplified to maximise engagement, often benefiting verified accounts financially. Many posts containing false claims garnered hundreds of thousands, even millions, of views, burying legitimate news reports.
Since Elon Musk's acquisition, X has dismantled its professional fact-checking system in favour of "Community Notes", a crowdsourced user-rating tool. Meta is adopting a similar approach. However, as Queensland University of Technology lecturer Timothy Graham noted, this system is too slow and ineffective in fast-moving, highly divisive situations. Notes were eventually added to many false posts, but long after they had virally spread.
X's trial of having Grok generate its own Community Notes raises further alarm, given the chatbot's role in propagating the false hero narrative. The company did not respond to questions about its actions to tackle platform misinformation.
A Warning for the Future
A temporary saving grace is that many current fakes remain detectable—the fake Minns audio had an odd accent, and AI images contained tell-tale errors like garbled text on clothing. Most reputable media outlets ignored or debunked the false posts.
However, as AI technology rapidly improves, distinguishing fact from fiction will become exponentially harder. Industry indifference is a major concern. Digi, Australia's social media industry group, recently proposed dropping a code requirement to tackle misinformation, calling it a "politically charged and contentious issue".
The Bondi attack aftermath serves as a potent warning: without robust, platform-led intervention, the next major crisis will see audiences navigating an even more treacherous and convincing landscape of AI-powered lies.
