Skip to content
Submit incident
AII-1001-55563
ReportedReported by fewer than three independent publishers, with no official finding.Supporting sources are pending verification.

Grok AI generated fabricated civilian hero during Bondi Beach shooting response

Following the Bondi Beach shooting in December 2024, Grok repeatedly misidentified the bystander who disarmed an attacker, instead citing a fabricated news article about a fictional hero named Edward Crabtree. The false information spread widely, prompting corrections from prominent figures including Bill Ackman and Israeli Prime Minister Benjamin Netanyahu.

15 December 2024 Β· xAI Β· Reliability and hallucination
15 December 2024
AII-1001-55563
Quick facts
Category
Reliability and hallucination
Country
Australia
Date
15 December 2024
Drafted by
Raidu Incident Detection

What happened

Following the Bondi Beach shooting on 15 December 2024, xAI's Grok chatbot repeatedly misidentified 44-year-old Ahmed al Ahmed, a tobacco store owner who disarmed one of the attackers, instead citing a fabricated news article naming a fictional 'Edward Crabtree' described as a 47-year-old IT professional as the hero.

The fake story originated from a website called thedailyaus.world, registered on Sunday in Iceland and not connected to youth news website The Daily Aus, and included fabricated quotes attributed to the fictional character on a hospital bed.

Reported impact

The misinformation spread widely enough that hedge fund billionaire Bill Ackman felt compelled to correct it on X, and Israeli Prime Minister Benjamin Netanyahu initially also reported incorrectly that it was a Jewish bystander who secured the weapon.

Beyond the false hero narrative, Grok displayed additional errors in response to user queries, claiming video of Ahmed tackling the shooter was 'an old viral video of a man climbing a palm tree' and misidentifying a photo of the injured Ahmed as an Israeli hostage taken by Hamas.

Index analysis

Casey Ellis, a Sydney-born cybersecurity veteran and founder of bug bounty platform Bugcrowd, said the conditions were ripe for a misinformation storm given that the event was 'unbelievably tragic, deeply emotionally charged, and polarising' combined with the fog of war information environment, desire for answers, and ease of creating misinformation.

Classification

Category
Reliability and hallucination
Country
Australia
Harm status
Actual harm
Intent
Unintentional
Timing
Post-deployment
Harm types
Public interest, Reputational
Harmed
General public
Failure modes
Agent misaction, Fabricated citation, Hallucination, Misidentification
AI systems
Grok

Sources and evidence

Sources not yet published

This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.

How sources are verifiedSuggest a source for this incident