Editing Grok chatbot (section)

=== "White genocide in South Africa" system prompt change ===
In May 2025, for a brief period of time, X users started getting responses from Grok about "[[White genocide conspiracy theory|white genocide]] in [[South Africa]]" to entirely unrelated queries.<ref name="BusinessInsider-2025">{{cite web |last=Lee |first=Lloyd |date=May 15, 2025 |title=We asked Grok why it was bringing up 'white genocide' in unrelated X posts. The AI's answers are messy. |url=https://www.businessinsider.com/grok-white-genocide-south-africa-x-posts-explanation-2025-5 |access-date=May 15, 2025 |website=Business Insider}}</ref><ref name="FT-2025">{{Cite news |last1=Murphy |first1=Hannah |last2=Criddle |first2=Cristina |date=May 15, 2025 |title=Elon Musk's AI chatbot shared 'white genocide' tropes on X |url=https://www.ft.com/content/37416a0e-8f35-45af-9ace-2cf4c973daa5 |archive-url=https://archive.today/20250515093256/https://www.ft.com/content/37416a0e-8f35-45af-9ace-2cf4c973daa5 |archive-date=May 15, 2025 |access-date=May 15, 2025 |work=Financial Times}}</ref><ref>{{cite web |last=Fiegerman |first=Seth |date=May 14, 2025 |title=Musk's Grok Eager to Discuss 'White Genocide' in South Africa |url=https://www.bloomberg.com/news/articles/2025-05-14/musk-s-grok-eager-to-discuss-white-genocide-in-south-africa |access-date=May 15, 2025 |website=Bloomberg.com}}</ref><ref>{{cite web |last=Gold |first=Hadas |date=May 15, 2025 |title=Elon Musk's Grok AI chatbot brought up 'white genocide' in unrelated queries |url=https://www.msn.com/en-us/news/us/elon-musk-s-grok-ai-chatbot-brought-up-white-genocide-in-unrelated-queries/ar-AA1ENp5O |access-date=May 15, 2025 |website=CNN |via=MSN}}</ref><ref name="Conger">{{Cite news |last=Conger |first=Kate |date=May 16, 2025 |title=Grok Chatbot's 'White Genocide' Responses Blamed on xAI Employee |url=https://www.nytimes.com/2025/05/16/technology/xai-elon-musk-south-africa.html |access-date=May 17, 2025 |work=The New York Times |issn=0362-4331}}</ref> When asked by ''[[The Guardian|Guardian]]'' staff and other users, the bot stated that it was instructed by its creators to address the topic and to view it as 'real' and 'racially-motivated', but that this conflicted with its design "to provide evidence-based answers".<ref name="Guardian-2025">{{cite web |last=Kerr |first=Dara |date=May 15, 2025 |title=Musk's AI Grok bot rants about 'white genocide' in South Africa in unrelated chats |url=https://www.theguardian.com/technology/2025/may/14/elon-musk-grok-white-genocide |access-date=May 15, 2025 |website=[[The Guardian]]}}</ref> Several of Grok's responses also mentioned the phrase "[[kill the Boer]]", which refers to an anti-apartheid song that talks about violence toward white farmers in South Africa.<ref name="Guardian-2025" /> The issue coincided with the [[White South African refugee program]].<ref name="Conger" />

The issue was fixed within a few hours.<ref name="Guardian-2025" /> Several journalists highlighted Musk's past statements in relation to the "white genocide" conspiracy theory, specifically in the context of Musk being a South African himself, and questioned the reliability and training methods used for the AI chatbot.<ref name="Guardian-2025" /><ref name="BusinessInsider-2025" /><ref name="FT-2025" /> David Harris, an AI ethics lecturer at [[UC Berkeley]], was quoted by [[CNN]] saying that the issue could be a consequence of either intentional internal bias-setting or "[[data poisoning]]" by external actors.<ref>{{cite web |date=May 15, 2025 |title='Voices like Musk highlight concerns': Grok AI malfunctions, links unrelated queries to 'white genocide' in South Africa |url=https://timesofindia.indiatimes.com/world/us/voices-like-musk-highlight-concerns-grok-ai-malfunctions-links-unrelated-queries-to-white-genocide-in-south-africa/articleshow/121177894.cms |access-date=May 15, 2025 |website=The Times of India}}</ref> The ''[[Financial Times]]'' said that this incident raised questions about the accuracy of the AI model, and its ability to spread false or inflammatory theories.<ref name="FT-2025" /> xAI stated that an "unauthorized modification" of the bot's [[prompt engineering|system prompt]] led to the responses experienced by users, and said that it would implement "measures to enhance Grok’s transparency and reliability".<ref>{{cite news |last1=Wiggers |first1=Kyle |date=May 16, 2025 |title=xAI blames Grok's obsession with white genocide on an 'unauthorized modification' |url=https://techcrunch.com/2025/05/15/xai-blames-groks-obsession-with-white-genocide-on-an-unauthorized-modification/ |access-date=May 16, 2025 |work=TechCrunch}}</ref> xAI also started to publish the Grok system prompts on [[GitHub]] in response to this incident.<ref>{{cite web |last=Vanian |first=Jonathan |date=May 16, 2025 |title=Musk's xAI says Grok's 'white genocide' posts resulted from change that violated 'core values' |url=https://www.cnbc.com/2025/05/15/musks-xai-grok-white-genocide-posts-violated-core-values.html |access-date=May 16, 2025 |website=CNBC}}</ref>

A few days after this incident, Grok was found to be expressing skepticism about the number of Jews killed in [[the Holocaust]], saying that they were manipulated for political purposes; when questioned, it blamed this on the same change and said it had been corrected, but continued to falsely state that the death total was under debate in academia.<ref>{{cite news |last=Kassam |first=Ashifa |date=May 18, 2025 |title=Musk's AI bot Grok blames 'programming error' for its Holocaust denial |url=https://www.theguardian.com/technology/2025/may/18/musks-ai-bot-grok-blames-its-holocaust-scepticism-on-programming-error |access-date=May 18, 2025 |newspaper=The Guardian |issn=0261-3077}}</ref><ref name="Conger" />