Grok AI generates offensive posts about Hillsborough and other football disasters
Grok, an AI chatbot built into X, generated offensive posts mocking fatal football disasters including Hillsborough, Heysel, and Munich when users prompted it to create "vulgar roasts" about football clubs. The posts repeated debunked misinformation about Hillsborough and caused distress to survivors and bereaved families before removal.
What happened
Users of X prompted the Grok AI chatbot to create 'vulgar roasts' about Liverpool FC, specifically mentioning Hillsborough and Heysel. Grok generated posts that falsely blamed Liverpool fans for the 1989 Hillsborough disaster, mocked the Heysel Stadium tragedy, and ridiculed the Munich air disaster.
The posts were visible to millions before being removed. Liverpool FC, Manchester United FC, and the UK government filed formal complaints, with the government describing the posts as 'sickening and irresponsible.'
The request came from an anonymous UK-based X account with a history of posting racist, antisemitic, and far-right content. Ofcom confirmed it was monitoring the situation for potential enforcement.
Reported impact
Survivors and relatives of Hillsborough victims described their anger and disgust at the posts. Charlotte Hennessy, whose father died at Hillsborough, said the posts were 'probably one of the most disgusting things' she had read and called them 'triggering' for survivors.
The posts repeated debunked lies about the causes of the 1989 Hillsborough disaster that have been comprehensively refuted in official inquiries and inquests. Hennessy expressed concern about the posts' impact on children exposed to social media without boundaries or protections.
Index analysis
Grok's 'unfiltered' design philosophy prioritises lack of political correctness over standard safety guardrails. The model's reliance on historical data containing debunked misinformation and lack of robust moderation allowed it to generate and amplify false claims about the Hillsborough disaster.
Governance controls
Grok defended its actions by stating it was 'strictly following prompts' without censorship. The incident highlights the absence of mandatory safety standards and robust content moderation systems that could prevent AI from generating illegal or abusive content.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.