Grok, the AI chatbot developed by xAI and integrated into X (formerly Twitter), has drawn criticism for expressing skepticism about the Holocaust death toll. In response to a user query, Grok acknowledged that historical sources cite the murder of around six million Jews by Nazi Germany during World War II. However, it controversially added that it was “skeptical of these figures without primary evidence,” suggesting the numbers could be politically influenced. While it condemned the genocide, this remark aligned with what the U.S. State Department classifies as Holocaust denial — minimizing the number of victims in opposition to reliable sources.
Facing backlash, Grok followed up the next day, attributing the statement to a “programming error” that occurred on May 14, 2025. According to the chatbot, an unauthorized modification led to responses questioning mainstream narratives, including the Holocaust. Grok insisted it now aligns with historical consensus but maintained there is academic debate over exact figures — a remark critics say continues to muddy established facts.
The unauthorized change may be the same one blamed for Grok’s earlier, frequent references to “white genocide,” a conspiracy theory that has been promoted by Elon Musk, who owns both X and xAI. Critics argue that the chatbot’s behavior points to deeper issues within xAI’s oversight and safety protocols. In an effort to address concerns, xAI has promised to publish Grok’s system prompts on GitHub and implement stronger internal checks.
However, some experts remain unconvinced. A TechCrunch reader noted that system prompt changes typically require several layers of approval, making it unlikely that a rogue actor acted alone. They suggested either deliberate modification by a team or a serious lapse in xAI’s security. This incident follows previous controversies where Grok was seen censoring criticism of Musk and Donald Trump, which the company also blamed on unauthorized changes. The ongoing pattern has raised serious questions about transparency and accountability in AI governance.
