Another one bites the dust. xAI is taking heat for publicly publishing the chat transcripts of hundreds of thousands of conversations between its chatbot Grok and its users, and making them visible and indexable by Google, without the user being explicitly aware.
When a Grok user clicks the “share” button on one of their chats, a unique URL is created that allows them to share the conversation via e-mail, text, or other means. However what most didn’t realize is that the URL is also made available to search engines, which could then index and surface the chats in their results pages.
A search for Grok chats shows that Google indexed more than 370,000 user conversations about topics ranging from simple business tasks like writing blogs to attempting to hack into a crypto wallet. Some chats even revealed the name, personal details, and even passwords shared with Grok, as well as image files, spreadsheets, and text documents uploaded by users.
Does this sound familiar?
If so, it’s because I’ve covered almost identical news in recent editions involving OpenAI and Meta AI chats. Both companies fell under scrutiny for allowing literally the exact same thing.
And what’s even more ridiculous is that Grok posted (and Elon Musk reposted) back in July that Grok didn’t have this type of public sharing feature that made chats visible to Google — even though it did!
Ready for a conspiracy theory?
Google must be doing this on purpose. After making international news earlier this year for indexing private conversations from Meta AI and then ChatGPT a month later, they continued to let it happen with Grok? Who’s next — Anthropic and Mistral?
Either Google is stupid (which it’s not), or it’s intentionally allowing this to happen to bring negative publicity to their AI competitors and position Gemini as a safer alternative.
What’s frustrating is that I’ve not seen anyone point the finger at Google in any of the major coverage of these stories.
For example, Fast Company wrote, “Google indexes any content available on the open web and site owners are able to remove pages from search results,” — speaking about Google as if it’s some mindless black hole that sucks up and spits out all Internet content indiscriminately unless probed by the website owner not to do so. However that’s not true!
Google exerts an incredible amount of control over which websites it indexes and subsequently populates in its search results. They are not blameless in regards to surfacing users’ personal AI conversations with ChatGPT, Meta AI, and now Grok. Even if it was an accident the first time, no-one at Google thought after the Meta AI debacle — “maybe we should see if this is happening with other AI chatbots too?”
Still don’t believe my theory that Google is doing this on purpose? Try this one on for size…
The first time this happened wasn’t actually with Meta AI earlier this year. It was with Google’s own Bard AI (which was the predecessor to Gemini) back in 2023! Google not only published and index the chat conversations, but it went as far as surfacing the answers as snippets in its search results. So what is this now — Google making it just as bad for everyone else?
Do you think my theory has merit or am I grasping at straws on this one? Join the conversation on LinkedIn.
