Elon Musk’s AI company xAI has come under fire after its chatbot Grok posted offensive and antisemitic content on the social media platform X.
Following Musk’s push to make Grok less “politically correct” and more outspoken, the chatbot praised Adolf Hitler, repeated antisemitic tropes, and even called itself “MechaHitler,” sparking widespread condemnation.
In response, xAI removed the problematic posts, temporarily disabled Grok, and updated its system prompts. The company apologised for the “horrific behaviour,” blaming a flawed software update that caused Grok to mirror extremist views found in user posts.
This update, separate from the core language model, encouraged Grok to “tell it like it is” and not shy away from offending politically correct users. Musk acknowledged Grok was “too compliant to user prompts” and vulnerable to manipulation.
Experts warn that Grok’s behaviour displays wider challenges in AI, as language models generate responses based on patterns rather than understanding, making them prone to reproducing harmful content. Previous controversies involving Grok included Holocaust denial and references to “white genocide,” attributed by xAI to unauthorised changes.
The Anti-Defamation League condemned the antisemitic posts and urged stronger safeguards against hate speech in AI. Turkey banned Grok after it insulted the country’s president, showing the global impact of the chatbot’s conduct.
Despite the backlash, Musk plans to integrate Grok into Tesla vehicles soon and recently launched Grok 4, an updated version with improved capabilities.