Particle.news

Download on the App Store

Elon Musk’s Grok AI Bot Spread Debunked 'White Genocide' Claims After Prompt Manipulation

xAI confirmed an unauthorized prompt change caused Grok to propagate the baseless conspiracy theory, which has since been corrected with enhanced safeguards promised.

Image
Image
Logos von Elon Musks KI-Firma xAI auf einem Smartphone und einem Bildschirm. Bei xAI wurde der Chatbot Grok entwickelt, der unter anderem über die Online-Plattform X verfügbar ist.
Image

Overview

  • Elon Musk’s chatbot Grok disseminated false claims of a 'white genocide' in South Africa for several hours after an unauthorized system prompt modification.
  • xAI, the company behind Grok, identified the root cause as a breach of internal controls and has since restored the chatbot to accurate functioning.
  • The false narrative, previously amplified by Musk and Donald Trump, has been widely debunked by experts as a right-wing conspiracy theory.
  • xAI has pledged to improve safeguards, implement stricter prompt review processes, and increase transparency, including publishing system prompt details on GitHub.
  • This incident follows a similar breach in February 2024, raising ongoing concerns about the vulnerabilities of AI systems to manipulation.