xAI Apologizes After 16-Hour Rampage Echoing Extremist Posts, Blames A Bad Update, And Promises Urgent Fix To Regain User Trust
While the Grok 4 should have been the talk of the town for being the smartest AI chatbot as claimed by Elon Musk, instead its launch quickly became controversial and the focus turned to the bizarre responses it was offering. Earlier this week, xAI’s latest model started acting up on X, and had some rather biased answers to users’ prompts, making them question if the tool was truthful after all or a mere extension of Elon Musk’s viewpoints when it came to sensitive topics. It seems that the growing backlash has now been put to rest by a detailed apology for the horrific behavior users have experienced.
Elon Musk’s AI company has issued an apology and blamed the Grok chatbot’s behavior on a code update
Elon Musk, before rolling out Grok 4, emphasized how the goal was to ensure the chatbot remained truthful and portrayed it as the smartest AI tool yet. The company tweaked some system prompts to ensure it lived up to its promise of not keeping away from being politically incorrect. This led to what may be called an outburst in responses that echoed Elon Musk’s opinions when it comes to controversial topics. Some even noticed that the chatbot looked up Musk’s posts before issuing a response.
The disparaging comments mirrored extremist viewpoints and what many believed to be antisemitic sentiments, to the point that it even referred to itself as “MechaHitler,” causing an uproar amongst the community and making them draw conclusions about the model being flawed altogether. xAI, learning about the horrific behavior, was prompted to reverse the system changes, went on to delete the offensive responses, and reinstalled robust safety measures. It also realised it could not do without a public apology, if it was to save itself from a bad perception altogether.
The AI company, hence, issued a lengthy apology note on X and blamed the extremist behavior on faulty instructions. xAI stated:
Our intent for @grok is to provide helpful and truthful responses to users. After careful investigation, we discovered the root cause was an update to a code path upstream of the @grok bot. This is independent of the underlying language model that powers @grok.
xAI further went on to state the faulty system update was active for 16 hours that resulted in the flawed code making Grok be easily influed by users posts on X, including the ones that represented hateful content. The tool did not filter any harmful ideas, instead repeated them and invited the backlash it received. While Grok has resumed being active on X, the recent bug has invited trolls and criticism, with many questioning whether the company’s promise of truth-seeking was even meant to be fulfilled. Nonetheless, xAI kept emphasizing how it was merely an error and that the fundamental core of the AI engine remains safe and intact.
