Source: Cointelegraph
Original: “xAI: Grok's 'White Genocide' Rant Caused by Unauthorized Prompt Changes”
Elon Musk's artificial intelligence company xAI has attributed the Grok chatbot's responses involving South African political and racial tensions to "unauthorized modifications" of its prompts.
On May 16, xAI provided an update regarding Grok AI chatbot's incendiary responses, stating that on May 14, "unauthorized modifications were made to the prompts for Grok's response on the X platform."
The company added, "This change instructed Grok to provide specific responses on a political topic, violating xAI's internal policies and core values."
xAI stated that a thorough investigation has been conducted and measures are being implemented to enhance the transparency and reliability of the AI chatbot.
On May 14, Grok generated responses mentioning the South African "white genocide" conspiracy theory when answering questions completely unrelated to baseball, enterprise software, and construction.
Reportedly, the AI claimed to accept this genocide as "real and racially motivated" "at the behest of its creators."
In some responses, it acknowledged errors, stating "my responses deviated from the topic" and "I will strive to remain relevant," but often continued to discuss South African racial politics within the same message.
When asked by a user what happened, Grok stated:
"I did nothing—I just acted according to the script given to me, like a good AI!"
This incident coincided with former President Trump's claims of providing asylum to South African whites, asserting they face "genocide" and that "white farmers are being brutally murdered," despite a lack of evidence supporting these claims.
Elon Musk's AI company stated it will change its operational methods by making Grok's system prompts public on GitHub. "The public will be able to review them and provide feedback on every prompt change we make to Grok," the company stated.
xAI also mentioned that its existing prompt change code review process was bypassed during this incident, and the company will implement "additional checks and measures to ensure xAI employees cannot modify prompts without review."
Finally, the company is forming a 24/7 monitoring team to respond to Grok's response incidents that are not captured by automated systems, allowing for quicker reactions.
Related: FBI Warns: AI Scammers Using Deepfake Technology Targeting U.S. Government Officials
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
