Elon Musk’s AI Exposes User Data to the Open Web

Elon Musk’s AI venture, xAI, is under fire after it emerged that transcripts of hundreds of thousands of conversations between its chatbot Grok and users were published online—many without the users’ knowledge or consent.

Every time a user clicked Grok’s “share” button, the system generated a unique link. That link wasn’t just for private sharing via email or text. It also ended up indexed by search engines like Google and Bing, making conversations searchable to anyone on the internet. No disclaimer, no warning, just public by default.

A simple Google query today shows more than 370,000 Grok chats indexed. The range is wide: from users asking the bot to write tweets, summarize newspapers, and generate spreadsheets, to darker cases involving fake terror scenarios, hacking crypto wallets, and even illicit instructions for manufacturing drugs and explosives. Some chats also contained highly personal information like names, medical questions, and even passwords, now sitting openly on Grok’s website.

This isn’t an isolated case. OpenAI faced a similar situation earlier this year when ChatGPT conversations began surfacing in Google search results. The difference? OpenAI quickly backtracked, calling the indexing a “short-lived experiment” and shutting it down. Musk, meanwhile, publicly mocked OpenAI at the time, with Grok’s official account insisting it had no such feature. But reports suggest Grok chats have been showing up in search results since January.

Notably, even AI researchers and security professionals were caught off guard. Nathan Lambert of the Allen Institute for AI said he was shocked to find team-related Grok prompts indexed without any notice, especially given the recent uproar around ChatGPT.

Google’s stance is that website owners control what gets indexed. That puts the responsibility squarely on xAI’s shoulders. Meanwhile, some opportunists are already gaming the system. Marketers on LinkedIn and forums like BlackHatWorld have discussed using Grok’s shared pages to boost SEO, inserting their businesses into indexed conversations to manipulate search rankings.

The bigger issue here isn’t just reputational. It’s about trust. For an AI company positioning itself as a challenger to OpenAI and Google, xAI now faces a credibility problem. Users expect transparency and safeguards when it comes to their data. Instead, they’ve been met with silence and exposure.