Hundreds of thousands of conversations with Grok exposed publicly on Google search results

Hundreds of thousands of user conversations with the artificial intelligence chatbot Grok – developed by billionaire Elon Musk – have suddenly appeared on Google search results, seemingly without the users' knowledge.

 

The problem stems from the sharing feature: when users click a button to send a chat transcript to someone else, the system generates a private link. But instead of sharing it only to the recipient, these links are accidentally indexed and publicly displayed on search engines.

As of Thursday, Google had indexed nearly 300,000 conversations with Grok. Forbes put the number even higher, at around 370,000. Some of the chats seen by the BBC show the bot being asked to create secure passwords, suggest weight-loss diet plans, or answer detailed medical questions. In some cases, users have tried to push the bot to its limits with sensitive requests. In one instance, Grok even provided detailed instructions on how to make Class A drugs in a lab.

 

The incident has many experts worried. Professor Luc Rocher, Oxford Internet Institute, warned that AI chatbots are becoming a 'privacy disaster in the making'. According to him, leaked conversations can reveal a lot of sensitive information about users, from names, locations, to mental health, business plans or even love affairs. Once they appear online, they will almost always be there.

Carissa Veliz, Associate Professor of Philosophy at the University of Oxford's Institute for Ethics in AI, also said it was 'very worrying' that users were not informed in advance that shared conversations would be made public on Google. 'Our technology doesn't even tell users how their data is being processed, and that's the problem,' she stressed.

This isn't the first time an AI chatbot has encountered similar problems. OpenAI experimented with allowing ChatGPT conversations to appear in search results if users shared them, before quickly stopping after a backlash. Meta has also been criticized for allowing chats from Meta AI to appear publicly in the app's 'discover' section.