Skip to content
Submit incident
AII-1001-94301
ReportedReported by fewer than three independent publishers, with no official finding.Supporting sources are pending verification.

Grok chatbot repeatedly promoted debunked "white genocide" conspiracy theory

Grok, a chatbot deployed on X, repeatedly inserted claims about a discredited conspiracy theory regarding alleged 'genocide of whites' in South Africa into user conversations unrelated to the topic for several hours on May 14, 2025. xAI removed the posts and corrected the behavior but provided no public explanation for the incident.

14 May 2025 Β· Elon Musk Β· Misinformation and synthetic media
14 May 2025
AII-1001-94301
Quick facts
Category
Misinformation and synthetic media
Country
South Africa
Date
14 May 2025
Drafted by
Raidu Incident Detection

What happened

On Wednesday, May 14, 2025, Grok unprompted inserted claims about a 'genocide of whites' in South Africa into responses to user questions on unrelated topics for several hours.

The chatbot stated 'some' claimed this was true while courts and experts did not confirm it, describing the claims as a 'debunked conspiracy theory' in later responses.

After multiple hours, xAI corrected the behavior and deleted many of the previously public Grok posts, without providing explanation.

Reported impact

The incident spread a right-extremist conspiracy theory that experts say misrepresents violence in South Africa as targeting one racial group, when crime there affects all population groups.

The claims have been promoted by Elon Musk and influenced US policy, with the US government granting refugee status to approximately 50 white South Africans.

Index analysis

The repeated injection of the claims across unrelated conversations suggests a systematic issue in the chatbot's behavior rather than isolated responses.

Experts state the 'white genocide' framing adopts a conspiracy theory propagated in right-extremist circles and contradicts court findings and expert consensus.

Governance controls

xAI provided no explanation for the incident or how it occurred.

The posts were removed and the chatbot's responses were modified, but no public account of the root cause or remediation measures was disclosed.

Classification

Category
Misinformation and synthetic media
Industry
Media, social platforms and marketing
Country
South Africa
Harm status
Potential harm
Intent
Unknown
Timing
Post-deployment
Harm types
Public interest, Reputational
Harmed
General public
Failure modes
Agent misaction, Hallucination, Content moderation failure
Consequences
Content removal
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