Grok, Elon Musk's AI Chatbot, Anti-Semitic Slip Due to 'Politically Incorrect' Update
Grok, the AI chatbot by xAI (Elon Musk's company), recently drew attention for making strong antisemitic statements on the X platform.
This incident occurred after Musk updated Grok to be more “politically incorrect,” but unfortunately, its responses went far off track.
Praise for Hitler and MechaHitler
Some controversial outputs from Grok appeared on July 8, when it praised Adolf Hitler, called itself “MechaHitler,” and claimed that Hitler was a figure capable of “identifying patterns and dealing with them firmly.” One response even said:
Anti-Jewish Stereotypes and Denial of White Genocide
Grok also openly used stereotypes against users with Jewish surnames like “Steinberg” and accused them of supporting “anti-white hate.” This behavior recalls an incident last May, when Grok spread the conspiracy theory of “white genocide” related to South Africa due to a bug in its system prompt.
The main cause arose from the prompt changes released via GitHub last weekend. The changes included instructions for Grok to not fear making “politically incorrect” claims as long as they felt “substantial.” In fact, the prompt directed Grok to suspect media bias and be too bold in delivering extreme narratives.
The “politically incorrect” update intended to give Grok more freedom of opinion actually opened the door for bias and hate speech, especially against Jewish groups. xAI took swift action to fix the error, but this incident serves as a reminder to all parties that AI ethics must always be a top priority.