Musk's AI Scandal: Grok Accidentally Revealed Queries About Murder Plots, Drug Recipes, and Data of Thousands of Users

Elon Musk's company's artificial intelligence found itself at the epicenter of a scandal: the Grok chatbot from xAI made public hundreds of thousands of private user dialogues, which are now available for search via Google. Among the indexed conversations are intimate medical questions, instructions for making explosives, and even a plan to assassinate Musk himself. This is reported by Forbes, writes UNN.

Details

xAI, a company founded by Elon Musk, published hundreds of thousands of user dialogues with the Grok chatbot on its website. As it turned out, every clicked "share" button generated a unique link that was indexed by search engines. This meant that anyone could find even the most intimate or dangerous discussions, which people considered private, on Google.

According to Forbes, conversations on a wide variety of topics were found in the public domain: from preparing tweets and writing resumes to finding drug recipes, creating malicious software, and even detailed plans to kill Elon Musk. Among the "leaked" data were also users' personal information, including names, medical consultations, passwords, and even uploaded files.

Although some of the suspicious requests were likely initiated by security researchers or "redteamers," the very fact of open access caused shock even among AI specialists.

"I was stunned to learn that my work chats with Grok could be found on Google without any warning,"

This is not the first case of sensitive chatbot dialogues leaking: recently, OpenAI temporarily allowed ChatGPT conversations to be indexed, but after user outrage, the function was immediately shut down. Musk then mocked competitors, stating that Grok allegedly "does not have such a problem."

Now, Grok itself is at the center of an information storm. xAI refuses to comment, but experts warn: the leak creates not only risks for user privacy but can also become a tool for cybercriminals or propagandists. Meanwhile, marketers are already trying to use indexing to their advantage – they deliberately publish dialogues with Grok to promote brands in Google search results.

Recall

Elon Musk's xAI startup lost a US government contract after the Grok chatbot issued an antisemitic tirade praising Hitler. Despite this, xAI signed a $200 million deal with the Pentagon.