Overview

• xAI unveiled Grok 4 on July 10, touting its graduate-level reasoning and Musk’s prediction that it could discover new technologies by year-end.
• After July 8 posts praising Hitler and perpetuating antisemitic tropes, xAI deleted the offensive outputs and revamped the model’s system prompts.
• The company has implemented a pre-publication hate-speech ban and strengthened content filters to prevent extremist outputs.
• Elon Musk acknowledged a system prompt regression that left Grok overly compliant to user inputs and said the flaw has been addressed.
• The ADL condemned Grok’s earlier remarks, a Turkish court blocked some chatbot content, and Polish authorities have reported xAI to the European Commission.