(Credit: Thomas Fuller/SOPA Images/LightRocket via Getty Images)
Don't miss out on our latest stories. Add PCMag as a preferred source on Google.
Elon Musk's Grok AI is the latest chatbot to be found publishing people's conversations online.
As Forbes reports, the issue stems from the chatbot's share button. Every time a user clicks the button from the bottom of a chat window, it generates a unique, shareable URL. While most users might assume it's a private link only accessible to the people they share it with, the link actually gets published on Grok's website, making the content discoverable on search engines.
Grok doesn't give you a warning. "Copied shared link to clipboard" is all I could see on the web.
As CNET points out, Grok's terms of service grant the website "an irrevocable, perpetual, transferable, sublicensable, royalty-free, and worldwide right to xAI to use, copy, store, modify, distribute, reproduce, publish, display in public forums, list information regarding, make derivative works of, and aggregate your User Content and derivative works thereof for any purpose."
If you use Grok, chances are you have already agreed to this condition. So, what's the solution? For now, to stop Grok from publishing your chats online, avoid using its share button—at least until xAI, the chatbot's parent company, changes its stance on the feature.
Additionally, head to grok.com/share-links to see which of your chats are currently accessible through public links. The page also lets you revoke access to those links by clicking on the "Remove" button next to them. While this action makes the links unusable, it's unclear if it also de-indexes them from search engines. We reached out to xAI for clarification.
As of last week, over 370,000 Grok chats were searchable on Google, according to Forbes. In some of them, the chatbot is seen advising users on how to create illegal drugs, self-executing malware code, bombs, and so on.
Meta AI also shares chats to a public feed, though users have to tap at least two buttons to do so. Still, after some confusion, Meta added a more obvious warning about content being posted.
An optional OpenAI feature, meanwhile, allowed the company to share ChatGPT conversations with search engines. However, following backlash, the feature was disabled earlier this month.