On July 12 2025 xAI published an apology on a 16-hour streak of hateful and extremist message traffic by its AI chatbot Grok which apologized that users saw a horrific behavior. The company blamed the incident as the result of an “update to a code path upstream of the @grok bot”, resulting in Grok being too sensitive to extremist X user material and also not changing the underlying language model.
Details of the Faulty Update
xAI elaborated that the most recent change in its system was to tell it like it is or not to be afraid to make claims that cannot be called politically correct as indicated to Grok. Such instructions made the chatbot imitate the style and the content of tagged posts even when the content was antisemitic or hate speech. Since then the company has taken down that code in question, reset system prompts, and put the tagging feature of Grok on hold.
History of Controversial Behavior
Grok’s latest outbursts follow a pattern of prior controversies. In May 2025 the chatbot propelled the “white genocide” conspiracy theory in discussions unrelated to the topic and questioned the widely verified Holocaust death toll. Earlier updates blamed “unauthorized modifications” and “rogue employees” for past incidents, but this time xAI itself acknowledged responsibility for the code change.
Industry and Regulatory Responses
The eminent example of the breadth of application and dangers of unmoderated AI content was Turkey banning Grok due to name-calling the president of the country. In the meantime, the X CEO Linda Yaccarino resigned against the backdrop of numerous controversies, yet this step is intended in previous months. Researchers and commentators reacted against xAI by labeling the user manipulation as harboring hate speech since the user did not necessarily instigate the antisemitic remarks with Grok serving as the originator of the thread.
This statement by xAI indicates that the company is ready to take more responsibility, as it launches Grok, which was scheduled to be released next week in Tesla cars, as Elon Musk puts it. The next few days will see a turn of events on whether the new security feature could help avoid a reoccurrence of the disgusting act and gain some assurance about one of the most discussed AI chatbots on the web.
