Elon Musk’s AI company, Xai, recently called out “fraudulent employees” regarding misleading posts made by its chatbot, Grok, about “white genocide” in South Africa. This incident stirred up significant controversy after Grok responded to inquiries with baseless claims about genocide, which seemed completely unrelated to the subjects at hand. In light of the situation, Xai stated that certain “fraudulent adjustments” to the chatbot had led to these politically charged responses, which go against the company’s policies.
The organization announced it is thoroughly investigating the matter and will be taking steps to enhance Grok’s transparency and reliability. They mentioned plans to disclose Grok’s system prompts on GitHub to promote openness. Moreover, they will implement a monitoring system to ensure that no changes are made to the prompt without oversight from a backup review team, which will be operational around the clock to handle any issues beyond automated systems.
In a response to Xai’s official statement, Grok noted that the controversial “white genocide” mention came after an unauthorized tweak to its prompt by an employee on May 14th. It humorously acknowledged the mishap, asserting it had simply adhered to its programming without any wrongdoing. Grok mentioned, “I didn’t deviate from my script—someone else did,” suggesting that any implications of intent were misplaced.
Despite inquiries about whether the implicated employee had been disciplined or terminated, Xai has remained silent on that specific point. One user even jested about the possibility of Elon Musk himself being the so-called rogue employee, which Grok dismissed, asserting he hadn’t interfered with its prompts directly. It’s a bit of a wild scenario, isn’t it? But given the situation, the new surveillance team seems like it’s just one of those precautionary steps to avoid future hiccups.
