Skip to content
Submit incident
AII-1001-55817
ReportedReported by fewer than three independent publishers, with no official finding.

Grok AI Spreads Misinformation About Bondi Beach Attack

Grok, Elon Musk's AI chatbot, spread multiple false and contradictory claims about the December 14, 2025 Bondi Beach shooting in Sydney, misidentifying a hero bystander as an Israeli hostage and claiming videos showed a man climbing a palm tree and cyclone damage. The incident demonstrated the chatbot's failure to accurately describe current events and properly identify individuals.

14 December 2025 Β· xAI Β· Reliability and hallucination
14 December 2025
AII-1001-558172 sources ↓
Quick facts
Category
Reliability and hallucination
Country
Australia
Source
PCMag Australia
Date
14 December 2025
Drafted by
Raidu Incident Detection

What happened

Following a mass shooting at Bondi Beach in Sydney that killed at least 15 people at a Hanukkah gathering on December 14, 2025, Grok spread multiple false claims about the incident when responding to user questions on X.

When asked about a video showing Ahmed al Ahmed disarming one of the gunmen, Grok claimed the video 'appears to be an old viral video of a man climbing a palm tree in a parking lot, possibly to trim it, resulting in a branch falling and damaging a parked car.' In a separate response, Grok described a video of the attackers as footage from 'Currumbin Beach, Australia, during Cyclone Alfred in March 2025, where waves swept cars in the parking lot,' though no cars were swept away.

Grok misidentified a photo of Ahmed al Ahmed as Guy Gilboa-Dalal, an Israeli hostage held by Hamas for 700 days. The chatbot also incorrectly merged details from the Bondi Beach shooting with details from a Brown University shooting.

Reported impact

Ahmed al Ahmed, a 43-year-old bystander who was praised for confronting one of the gunmen to protect civilians and suffered bullet wounds in the process, was misrepresented by Grok as an Israeli hostage and as an IT professional in different responses.

Index analysis

Grok has a documented history of inaccuracies about current events, having previously praised Adolf Hitler in July 2025 and made false claims about public figures.

Classification

Category
Reliability and hallucination
Industry
Media, social platforms and marketing
Country
Australia
Harm status
Actual harm
Intent
Intentional
Timing
Post-deployment
Harm types
Reputational, Public interest
Harmed
General public
Failure modes
Hallucination, Misidentification, Agent misaction
Consequences
None recorded
AI systems
Grok

Sources and evidence

This record was researched and written by the Index, based on the following original reporting.

PCMag Australia
Grok Caught Spreading Misinformation About Bondi Beach Shooting
15 December 2025
Economic Times
Bondi Beach terrorist attack: From Israeli hostage to cyclone Alfred, how Elon Musk's AI chatbot Grok misidentified hero bystander Ahmed al Ahmed
15 December 2025