According to an exclusive Fast Company article, Google is indexing shared conversations from ChatGPT, potentially exposing private exchanges to a wider audience. This development raises concerns about the privacy of users who share their chats with others, assuming a limited scope of visibility.
By using a specific Google site search that includes a portion of the link generated when a user clicks the “Share” button within ChatGPT, it is possible to uncover conversations containing sensitive personal information. These details include accounts of addiction struggles, experiences with physical abuse, discussions of mental health issues, and even concerns about AI surveillance. The conversations are accessible even though ChatGPT does not explicitly display user identities. However, some users may inadvertently reveal identifying information by sharing specific personal details during their interactions with the chatbot.
The “Share” function is typically used when sending a conversation to a friend via platforms like WhatsApp or saving the URL for personal reference. The expectation is that this action maintains a limited audience. Many users are likely unaware that sharing their conversations in this way can result in their appearance in Google search results, making them accessible to a much broader audience. The extent to which affected users understand their conversations are now publicly accessible remains unclear.
A Google site search reveals nearly 4,500 conversations. While not all of these contain sensitive or identifying information, the figure may not represent the total number of indexed conversations, as Google may not have indexed all shared chats. Due to the personal and sensitive nature of these conversations, Fast Company has chosen not to provide links or detailed descriptions of the exposed chats.
