Liv McMahon Technology Reporter
Getty ImagesHundreds of thousands of user conversations with Elon Musk’s artificial intelligence (AI) chatbot Grok have been revealed in search engine results – seemingly without the users’ knowledge.
Unique links are created when grok users press the button to share a transcript of their conversation, but as well as sharing the chat with the intended recipient, the button also appears to have made the chat searchable online.
Google Search revealed on Thursday that it had indexed nearly 300,000 grok conversations.
This has led one expert to describe AI chatbots as a “privacy disaster in progress.”
The BBC has reached out to X for comments.
The appearance of GROK chat in search engine results was first reported by tech industry publication Forbes, which counted more than 370,000 user conversations on Google.
Among the chat transcripts seen by the BBC were examples of Musk’s chatbot being asked to create a secure password, provide meal plans for weight loss and answer detailed questions about health conditions.
Some indexed transcripts also showed users attempting to test the boundaries of what Grok would say or do.
In one example seen by the BBC, a chatbot provided detailed instructions on how to make a Class A drug in a laboratory.
This is not the first time that people’s conversations with AI chatbots have appeared to be broader than initially realized through “sharing” features.
Openai recently ran an “experiment” where ChatGpt conversations were seen in search engine results when shared by users.
A spokesperson told BBC News at the time that it was “testing ways to make it easier to share useful conversations while keeping users in control.”
They said that user chats were private by default and users had to explicitly opt in to sharing them.
Earlier this year, Meta faced criticism after shared user conversations with its chatbot Meta AI appeared in a public “discover” feed on its Local app.
“Privacy disaster”
While user account information may be anonymized or obscured in shared chat metrics, their prompt may still contain and risk revealing personal, sensitive information about someone.
Experts say this raises major concerns about user privacy.
“AI chatbots are a progressive privacy disaster,” Professor Luc Rocher, associate professor at the Oxford Internet Institute, told the BBC.
They said “leaked conversations” from chatbots have exposed users’ information, ranging from full names and locations to sensitive details about their mental health, business operations or relationships.
“Once leaked online, these conversations will stay there forever,” they added.
Meanwhile, Carissa Veliz, professor of AI philosophy at the University of Oxford’s Institute for Ethics, said that users not being told that shared conversations would appear in search results is “problematic.”
“Our technology won’t even tell us what it’s doing with our data, and that’s the problem,” she said.
