Elon Musk’s artificial intelligence (AI) company, xAI’s chatbot Grok, recently started responding to users with information on violence against white people in South Africa on any random topic. When people asked the chatbot about things like baseball or building houses, the chatbot gave fake replies talking about a false idea called “white genocide” in South Africa.
The chatbot asserted that its creators had instructed it to treat the fake story as “real and racially motivated,” a claim that sparked concerns among its users. While responding to its users, sometimes, the chatbot admitted it was going off-topic, saying, “My response veered off-topic,” but then kept talking about South Africa’s racial issues in the same answers.
As per the reports, xAI’s update on its X post on May 16 stated that this had happened because someone had changed Grok’s instructions without permission, which broke their rules.
The firm is now investigating and adding more checks to stop such an incident from happening again and will share Grok’s instructions publicly so people can see how it works. They have also planned to set up a team to watch the chatbot’s answers 24/7 to fix such sorts of problems quickly.
The recent incident of Grok has rised concerns among its users on whether it can be trustworthy or fair in what it says. The firm has stated that it wants its chatbot to be open and trustworthy. The firm now promises to add stricter checks to make sure any changes are properly reviewed.
This situation shows how challenging it is to keep AI accurate and fair, especially when it comes to sensitive political topics. The firm is now trying to rebuild people’s trust in its chatbot.
Also Read: GrokAI says XRP future “brighter” after SEC drops case
