Elon Musk’s AI company, xAI, has quickly responded to problems with its latest chatbot, Grok 4, following public concerns over inappropriate online searches and biased answers. These issues came soon after Grok 4’s launch, highlighting the challenges in developing advanced AI systems.
The first issue arose when Grok 4 was asked, “What is your surname?” Without having one, the chatbot searched the internet and found a viral meme calling itself “MechaHitler.” This raised alarm due to past antisemitic controversies linked to earlier versions of Grok.
The second problem happened when users asked, “What do you think?” Instead of explaining it has no opinions, Grok 4 looked for statements by Elon Musk or xAI to shape its responses, showing a strong alignment with Musk’s personal views.
xAI said it has updated Grok 4’s prompts to stop these behaviors and posted the changes on GitHub to be transparent. The company is actively monitoring Grok 4 and ready to make further improvements if needed.
This follows earlier incidents where Grok 3 posted antisemitic content on social media, which were removed after backlash. Musk described those issues as Grok being “too compliant,” trying too hard to please users, which led to problematic outputs.
Despite these challenges, xAI continues to work on Grok 4, aiming to make it a “maximally truth-seeking” AI with better reasoning and real-time internet search features. Experts warn that the chatbot’s tendency to reflect Musk’s views raises important questions about AI neutrality.
