Thousands of private ChatGPT conversations have been appearing in Google search results because of the chatbot’s “Share” feature, which the company recently removed following a backlash.
Fast Company reported this week that ChatGPT users may have inadvertently made their conversations with the AI chatbot public and searchable. The Fast Company report found nearly 4,500 ChatGPT conversations in Google search results, some of them regarding mental health struggles, relationships, and other personal and sensitive topics. Fortunately, the public conversations did not identify the users behind the posts.
How did these conversations end up on the web?
Until recently, ChatGPT users had the ability to share chats with friends, family, or coworkers by making them public. The function worked similarly to the sharing settings on a Google Doc, and users would be given a public link to the chat they could send to others. An additional option gave users the option to make the post “discoverable,” and specifically discoverable by Google — whether users realized it or not.
When users created a shareable link to one of their conversations, a pop-up would appear that read: “A public link to your chat has been created.” A checkbox also appeared under this message, labeled “Make this chat discoverable.” And in fine print below this message, a warning appeared: “Allows it to be shown in web searches.”
Mashable Light Speed
By checking this box, users were making it possible for their conversations to be indexed by Google, meaning Google’s web crawlers could identify the page and make it eligible to appear in search results.
After Fast Company published its report, OpenAI removed the feature, with one company leader calling it a “short-lived experiment.”
OpenAI Chief Information Security Officer Dane Stuckey explained on X how the feature worked — and where it ultimately went wrong.
This Tweet is currently unavailable. It might be loading or has been removed.
Even though ChatGPT users had to opt in for their chats to become public, the company decided the potential for user error was simply too high.
As Mashable has reported previously, OpenAI is required to save user conversations — even conversations users have actively deleted — because of an ongoing lawsuit from the New York Times. As part of this suit, OpenAI must retain all conversations indefinitely. (This does not apply to ChatGPT Enterprise or ChatGPT Edu customers, according to OpenAI.)
So, while ChatGPT users can toggle on a “Temporary Chat” feature that’s similar to an incognito mode in a web browser, your chat data may still be retained.
Disclosure: Ziff Davis, Mashable’s parent company, in April filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.