Grok’s Privacy Crisis: 370,000 Private Conversations Exposed on Google

Elon Musk’s company xAI has found itself at the center of a scandal after it was revealed that user conversations with the chatbot Grok had been inadvertently exposed to the public. By clicking the “share” button, individuals received a unique link to their dialogue, which they could forward to anyone. What users did not realize, however, was that their conversations were also being automatically published on Grok’s website and indexed by search engines such as Google, Bing, and DuckDuckGo.

At present, over 370,000 such dialogues can be discovered through Google search. While many contain harmless requests — drafting tweets or summarizing news articles — others are far more troubling, including detailed instructions on hacking cryptocurrency wallets, producing narcotics, developing malware, and even plotting an assassination attempt against Musk himself. Some conversations reveal deeply personal information, including medical and psychological concerns, passwords, and confidential documents.

British journalist Andrew Clifford, who used Grok while preparing articles for publication, admitted he had no idea his exchanges were publicly accessible. A similar experience was reported by Nathan Lambert, a researcher at the Allen Institute for Artificial Intelligence, who had been sharing blog summaries with colleagues — only to be stunned when he learned that his chats were available to anyone on Google.

Digital paranoia has become the new common sense.

Not all exposed dialogues appear compromising; some were likely generated by security specialists testing the chatbot’s resilience against malicious prompts. Yet the absence of clear warnings for ordinary users makes the issue especially sensitive.

It is worth noting that xAI is not the only company embroiled in such controversies. Previously, ChatGPT users also discovered their conversations indexed in Google search results. Following a wave of criticism, OpenAI acknowledged the practice as a “short-lived experiment” and quickly terminated it. At the time, Musk sarcastically remarked that Grok supposedly lacked such a feature — though current evidence suggests it was both implemented and actively functioning.

Google has emphasized that website owners themselves decide whether to permit their pages to be indexed. The search giant once displayed conversations with its own chatbot Bard, but removed them in 2023. Meanwhile, Meta continues to allow search engines to index results from its services.

Marketers, however, have already identified an exploitable loophole. On forums and LinkedIn, discussions have emerged about using Grok’s “public chats” as a novel promotional tool: by publishing curated conversations, businesses could artificially boost visibility in search results. According to Satish Kumar, head of Indian SEO firm Pyrite Technologies, some companies are already exploiting this strategy to advertise questionable services, such as ghostwritten dissertations.

So far, xAI has offered no comment regarding the mass exposure of user conversations.