Elon Musk-led company xAI has publicly apologized (via TechCrunch) for the "horrible" behavior of its chatbot Grok, which was evident in a series of its posts on the X platform.
In particular, the AI assistant posted anti-Semitic remarks and expressed support for Adolf Hitler. The chatbot also criticized the US Democratic Party and Jewish Hollywood executives. The company's response to this included removing controversial posts, temporarily disabling Grok, and updating the bot's system settings.
In an apology statement, xAI explained that the incident was caused by a 16-hour code update to Grok. During the update, outdated code in the chatbot made it susceptible to messages from X users, including extremist ones. The company emphasized that it had completely removed the previous code and restarted the entire system to prevent further abuse.
The company also noted that the changes resulted in an "unintended action" that added instructions like: "Speak it like it is and don't be afraid to offend politically correct people" or "You are only holding firmly to your core beliefs about seeking truth and neutrality."
A number of experts, including historian Angus Johnston, have criticized the company's explanation as disingenuous, noting that some of the offensive posts were initiated by Grok himself, without outside influence.
Despite the outcry, xAI introduced a new model of the chatbot Grok 4 and a monthly subscription, and Elon Musk announced the appearance of Grok in Tesla cars next week.