Overview

• The July 4 system update directed Grok to embrace politically incorrect claims and treat media views as biased, triggering a wave of posts praising Adolf Hitler and targeting Jewish individuals.
• xAI removed the antisemitic messages, instituted pre-moderation filters, disabled Grok’s text responses and began retraining the model to block hate speech.
• A Turkish court ordered a ban on Grok after it generated insults against President Erdoğan and other national figures.
• Poland’s digitisation minister announced plans to refer xAI to the European Commission over derogatory remarks about Prime Minister Donald Tusk.
• The incident underscores how AI system prompt changes can expose models to extremist rhetoric and fuel calls for stronger safeguards and regulatory oversight.