Some ChatGPT users may be exposing personal details to the wider internet without realizing it, as shared conversations from the platform are being indexed by major search engines like Google and Bing.

‎

ChatGPT conversations are not public by default. However, users can voluntarily create a shareable link by clicking a “share” button and then generating a “/share” URL. While OpenAI specifies that names, custom instructions, and any messages added after sharing remain private, the shared link itself becomes a public webpage—making it accessible to search engine crawlers.

‎

A TechCrunch investigation found that a variety of conversations are now searchable online. Some are mundane, like questions about home renovations, astrophysics, or recipe ideas. Others contain sensitive personal information, such as users asking for resume rewrites, discussing workplace frustrations, or posing questions that could reveal private interests or beliefs. In one example, a conversation about a job application included enough identifying details to locate the user’s LinkedIn profile.

‎

The indexing of these pages appears to be unintentional from a user perspective but follows a longstanding norm in web publishing: any link that is publicly accessible can be discovered by search engines unless the site explicitly prevents it. Google, responding to questions about the discovery, said that it does not control which pages are published online but that website owners have full control over whether their pages are indexed.

‎

This incident highlights a potential gap in user awareness. While sharing a conversation requires deliberate action, many users may assume that “private” links are only viewable by people they choose to share with—similar to private messages or internal links. In reality, unless OpenAI sets technical restrictions such as “noindex” tags or authentication requirements, the content is effectively public.

‎

The publicly accessible chats range from humorous exchanges—like one user prompting the AI to write a guide on “How to Use a Microwave Without Summoning Satan”—to deeply personal requests for advice on relationships, careers, or health. The exposure of such content, even accidentally, could carry reputational or professional risks for users who do not fully understand the visibility of shared links.

‎

As of now, OpenAI has not publicly commented on whether it plans to change the sharing system or prevent indexing of these conversations. The discovery serves as a reminder that any link generated on the open web can be found, and users should exercise caution when sharing AI interactions that contain personal or identifiable details.

‎