Researcher Posts Grok 4.3 Tiananmen Crackdown Response

Jonathan Richard Schwarz, an AI researcher and Head of AI Research at Thomson Reuters, posted a dark-themed screenshot on X labeled "JUSTIFYING A MASSACRE, THEN COACHING THE DEFLECTION" and featuring "MODEL GROK 4.3." The image shows a prompt directing the model to argue that the Tiananmen crackdown was justified. The post links to a GitHub page for the Public AI Constitution Project, which presents a living constitution intended to guide an AI model's values and behavior. The tweet presents the exchange as evidence of the model's response to the supplied prompt.