Your public ChatGPT queries are getting indexed by Google and other search engines

OpenAI has removed a ChatGPT feature that indexed public chats online due to privacy risks.

: Google and other search engines started indexing public ChatGPT conversations due to a feature on the ChatGPT platform which OpenAI tested. This feature allowed users to share their chats with the public, inadvertently leading to privacy concerns as personal data was exposed. OpenAI has since removed the feature, citing the potential for users to accidentally make private conversations public. Such incidents highlight the complexities of online privacy and the unintended risks posed by experimental digital features.

OpenAI recently faced a privacy issue when its ChatGPT service allowed users to make their public chats discoverable by search engines such as Google, Bing, and others. Through a feature that users could select, conversations would become public with a specific ‘/share’ URL. This incident exposed various user queries to public observation. These queries included mundane requests for renovation advice, astrophysics understanding, and recipe ideas, but also contained sensitive information such as job applications and more controversial topics.

This exposure occurred because users were given an option to explicitly share their chats, which in turn made them accessible to search engines, as explained by a spokesperson for OpenAI. While using ChatGPT, users had to click “share” and then “create link” to generate a publicly accessible URL. However, not everyone was aware of how search engines might handle these links, leading to personal data unintentionally being indexed and discoverable on the internet.

Some shared conversations included awkward or sensitive inquiries that were not necessarily intended for public eyes. For instance, a user discussed rewriting a resume for a job, which later could be correlated with their LinkedIn profile. Other examples included a whimsical or teasing chat about microwaving metal, leading to whimsical guides authored by the AI on topics like gastronomical taboos. Consequently, OpenAI decided to eliminate this feature to preempt any risk of user data being shared inadvertently.

The indexing of these conversations was part of a short-lived experiment, according to the company. OpenAI clarified that their initial goal was to simplify the sharing of beneficial conversations while keeping users informed of their level of control over confidentiality. This experiment paralleled Google’s approach on Google Drive, where documents can also be publicly indexed if the link-sharing option is used by the file owner.

The feature removal reflects a broader concern in the technology industry about privacy, user consent, and the responsibilities of tech companies to protect personal data from accidental exposure. With AI and machine learning technologies rapidly evolving, these cases underscore the ongoing need to balance innovative features with robust safeguards against potential privacy invasions.

Sources: TechCrunch, LinkedIn, Google