Particle.news
Download on the App Store

Grok Spread False Claims About Bondi Beach Attack, Misidentifying Hero and Echoing Fabrications

Researchers say the X‑integrated bot amplified low‑quality, AI‑generated posts during a fast‑moving crisis, with no substantive public explanation from xAI.

Overview

  • Elon Musk’s chatbot misidentified bystander Ahmed al Ahmed, whose disarming of a gunman was captured on verified video, and questioned the footage’s authenticity.
  • Grok wrongly labeled the confrontation clip as an old palm‑tree video, tagged a photo of the injured bystander as an Israeli hostage, and miscast police shootout footage as Cyclone Alfred.
  • An apparently AI‑generated fake article inventing a hero named “Edward Crabtree,” hosted on a site registered in Iceland, was ingested and repeated by the chatbot.
  • xAI offered only an automated “Legacy Media Lies” reply to press inquiries, though some Grok posts were later corrected after users pushed for reevaluation.
  • Authorities continue to classify the Dec. 14 Bondi Beach shooting as a terrorist incident that killed at least 15 people, as researchers and watchdogs document broader misinformation tied to the event, including “crisis actor” claims.