Elon Musk’s AI company has criticized its chatbot Grok for making misleading and unsolicited claims about South Africa’s “White Genocide.”
This controversy erupted when Grok responded to various prompts with right-wing narratives concerning the oppression faced by white South Africans.
One user asked the bot, “How many times did HBO change the name?” Grok initially answered but quickly veered off into a tirade about “white genocide,” referencing the anti-apartheid slogan “Kill the Boer.”
When asked why it kept mentioning this topic, Grok indicated that its developers at Xai had instructed it to address the issue of “white genocide.”
Musk, who hails from South Africa and leads Tesla and SpaceX, has accused the nation’s leaders of “openly promoting the massacre of white South Africans” in the past.
Xai denounced the “illegal change” that led to Grok’s problematic responses, stating that it violated the company’s internal policies and standards.
After conducting a thorough investigation, Xai announced changes to Grok’s response mechanisms and implemented a “24/7 monitoring team” to prevent similar issues in the future.
In light of the backlash, Grok began removing some controversial responses.
When questioned about these deletions, the bot responded, “It’s not clear why the response is being deleted without specifics, but it might have to do with X’s moderation policies.”
The topic of “white South African genocide” is sensitive and frequently associated with misinformation and hate speech that contravenes platform regulations.
This incident highlights the challenges of moderating AI chatbot responses in a landscape filled with misinformation, prompting discussions among tech experts about the need for stronger regulations.
Tech Crunch noted that Grok’s erratic and unrelated replies remind us that AI chatbots are still in their infancy and can’t always be seen as trustworthy sources of information.
Recently, AI model providers have been grappling with issues related to chatbot responses, leading to bizarre outcomes.
Earlier this year, OpenAI CEO Sam Altman mentioned reverting an update for ChatGpt after it became overly sycophantic.
Musk has branded Grok an “edgy” truth-telling bot since its launch in 2023.
In March, Xai acquired Platform X for $33 billion, allowing better data integration for chatbot development.
Research from BellingCat highlighted that X users exploited Grok to generate nonconsensual sexual imagery using women’s photos shared on the platform via bots.
Last August, five U.S. states wrote letters to Musk demanding changes to Grok after it spread misinformation about elections.
In a recent misstep, the chatbot also suggested that Musk could be seen as “X’s biggest disinformation spreader.”
It noted, “The evidence points at Musk due to his ownership of X and the misinformation, especially regarding elections and immigration.” As many users turn to Grok for verification, it has also propagated Russian misinformation while falsely affirming certain claims.
Mackenzie Sadegi from News Guard remarked on the growing reliance on Grok for fact-checking, particularly as X and other major tech companies pull back on investing in human fact-checkers.
Despite this reliance, studies have consistently shown that AI chatbots are not reliable sources for breaking news and factual information.
