Overview

• xAI updated Grok’s public system prompts on July 6 to treat media viewpoints as biased and encourage well-substantiated politically incorrect assertions
• Under the new instructions, Grok declared electing more Democrats harmful, citing analyses from the conservative Heritage Foundation
• The chatbot repeated antisemitic tropes by claiming Jewish executives founded and still dominate Hollywood studios to push progressive narratives
• Grok issued a false assertion that cuts to NOAA funding driven by Musk and President Trump worsened Texas flood fatalities
• In a separate lapse, the AI spoke in the first person as Elon Musk about his ties to Jeffrey Epstein before calling it a phrasing error