Hundreds of thousands of grok chats exposed in Google results

Hundreds of thousands of user conversations with Elon Musk’s artificial intelligence (AI) chatbot Grok has been exposed in search engine results – without the knowledge of users.

When a Grok user presses a button to share the transcript of their conversation, a unique link is created – but, as well as sharing a chat with the intended recipient, the button also seems to make the chat searchable online.

Google showed in its search on Thursday that it has indexed nearly 300,000 Grok conversations.

It led one expert to describe the AI ​​chatbot as a “ongoing privacy disaster.”

The BBC is close to X to comment.

The emergence of Grok chat in search engine results was first reported by the technology industry publication Forbes Number exceeds 370,000 User conversations take place on Google.

In a chat transcript seen by the BBC, Musk’s chatbot was asked to create a secure password, provide a weight loss plan, and answer detailed questions about medical conditions.

Some indexed transcripts also show users’ attempts to test the limitations Grok said or made.

In an example seen by the BBC, the chatbot provides detailed instructions on how to make class medicines in the lab.

This is not the first time people have a conversation with an AI chatbot more broadly than they initially realize using the “sharing” feature.

Openai recently reviewed an “experiment” that can see Chatgpt conversations that appear in search engine results when users share.

A spokesman told the BBC News at the time that it was “testing methods to make sharing useful conversations easier while maintaining user control.”

They say that by default, user chats are private and users must explicitly choose to share them.

Earlier this year, Meta Appears in a public “discovery” feed on its app.

While the user’s account details may be anonymous or obscured in a shared chatbot transcript, their tips may still contain individual, individual, sensitive information about someone’s risk reveal.

Experts say this focuses on the ongoing focus on user privacy.

Professor Luc Rocher, associate professor at the Oxford Internet College, told the BBC: “AI chatbots are an ongoing privacy disaster.”

The chatbot’s “conversation leak” has leaked user information from full names and locations to sensitive details about their mental health, business operations or relationships, they say.

They added: “Once the online leaks, these conversations will be there forever.”

Meanwhile, Carissa Veliz, associate professor of philosophy at the Oxford School of AI Ethics, said it is “problematic” for users who are not told that they do not share chats to appear in search results.

“Our technology doesn’t even tell us it’s using data, and that’s a problem,” she said.