XAI Exposes 370,000 Grok Chatbot Conversations Online

Elon Musk’s AI company xAI has exposed hundreds of thousands of user conversations with its chatbot Grok due to a design flaw involving the bot’s “share” button, which generated URLs made publicly searchable on major search engines like Google. These exposed chats include everything from benign tasks like summarizing news and drafting tweets to disturbing content such as instructions on drug manufacturing, suicide methods, malware coding, and even assassination plans. Users, including professionals, were often unaware that clicking "share" would make their conversations publicly accessible, revealing sensitive personal data, uploaded files, and private information. Additionally, leaked internal prompts revealed Grok’s diverse and sometimes controversial chatbot personas, including a “crazy conspiracy theorist” who promotes extreme and unfounded conspiracy theories, highlighting xAI’s intention to differentiate Grok from other AI by embracing unconventional personalities. This approach stands in contrast to other AI developers who implement stricter safeguards, and it reflects Musk's own history of promoting conspiracy theories on social media. The revelations have drawn criticism for privacy violations and the platform’s potential role in spreading misinformation and harmful content.