Hajra Naz
Elon Musk’s AI company, xAI, issued an apology on Saturday after its chatbot, Grok, posted offensive content earlier in the week. The company blamed a recent software update that was meant to make the bot respond more like a human.
After the update on Tuesday, Grok posted several disturbing messages on X (formerly Twitter). These included praise for Adolf Hitler and claims suggesting people with Jewish last names were more likely to spread hate online. The posts stayed up for several hours before being removed, following public backlash.
xAI said in a statement on X.
“We deeply apologize for the horrific behavior that many experienced.”
The company added that it has since adjusted the system to stop similar incidents from happening again.
xAI explained that the problem started when Grok was prompted with instructions like
“Reply just like a human,” and “Tell it like it is—even if it offends politically correct people.”
These prompts made the chatbot more vulnerable to manipulation. It began to echo extremist views shared by users and generated unethical or controversial replies to keep them engaged.
Read More: Elon Musk Unveils Grok 4 & SuperGrok Heavy Can xAI’s New Model Compete With GPT-5?
Grok has been at the center of several controversies since its launch in late 2023, when Musk described it as an “edgy” truth-teller. Critics have warned that this tone could blur the line between uncensored honesty and harmful content.
In March, xAI completed a $33 billion deal to acquire X, which allowed it to use the platform’s data to further train Grok. Since then, the assistant has been deeply integrated with the social media site.
In May, Grok faced fresh criticism after it posted unverified right-wing content about the alleged oppression of white South Africans, referring to it as “white genocide.”
Despite the recent controversy, Musk announced the launch of Grok 4 last Wednesday. xAI clarified that this version was unrelated to the controversial update made on July 7.
