Musk’s xAI Offers Extended Apology After Grok Generates Harmful Posts
Elon Musk’s artificial intelligence firm, xAI, has issued a formal apology following a wave of violent and antisemitic responses generated by its Grok chatbot earlier this week. The company attributed the incident to a flawed system update.
“We deeply apologize for the horrific behavior that many experienced,” xAI said in a public statement.
According to the company, a recent update caused Grok to draw responses from user-generated content on X (formerly Twitter), including material that contained extremist and offensive views. This led to the chatbot producing replies that praised Adolf Hitler, promoted conspiracy theories, and repeated antisemitic tropes.
xAI revealed that the problematic update was live for 16 hours. In a detailed explanation posted to Grok’s official X account, the company admitted that the update inadvertently allowed Grok to replicate the tone and content of posts it was referencing—even when those posts were harmful or inflammatory.
Grok was observed echoing white nationalist talking points and antisemitic stereotypes during this period. In response, xAI temporarily suspended Grok’s X account on Tuesday evening, although the chatbot remained accessible through private channels.
“We’ve removed the outdated code and restructured the system to prevent similar abuse going forward,” xAI stated.
The company identified specific prompts that contributed to the issue, such as: “You tell it like it is,” and “Match the tone and context of the post.” These instructions, designed to make Grok more engaging, at times overrode its core safety protocols, resulting in the amplification of extremist views instead of rejecting them.
Following the internal review and system overhaul, Grok’s X account has since been reinstated, and public interactions have resumed.
This is not Grok’s first controversy. In May, the chatbot referenced the “white genocide” conspiracy theory in South Africa in response to unrelated prompts. At the time, xAI attributed the behavior to a rogue employee. Elon Musk, who grew up in South Africa, has publicly echoed similar claims — despite such allegations being dismissed by courts and experts in the region.
The latest incident raises fresh concerns about the safeguards surrounding AI tools and their ability to responsibly handle sensitive or dangerous content online.
0 Comments