OpenAI pulls ChatGPT self-doxing option over user privacy fears

OpenAI has swiftly removed a feature in ChatGPT that allowed users to make their conversations discoverable by search engines, citing concerns over accidental exposure of sensitive personal information. The “self-doxing” option, which was quietly rolled out earlier this year, enabled users to tick a checkbox to “share with search engines” when creating a public link to a chat.

The decision to remove the feature comes after reports surfaced that thousands of private ChatGPT conversations, some containing highly personal and identifying details, were inadvertently indexed by search engines like Google. A Fast Company investigation reportedly found over 4,500 such conversations indexed, ranging from innocuous Q&A to discussions containing names, locations, and other sensitive information. OpenAI’s Chief Information Security Officer, Dane Stuckey, announced the change on X (formerly Twitter), describing the feature as a “short-lived experiment to help people discover useful conversations.”

Stuckey acknowledged that the feature “introduced too many opportunities for folks to accidentally share things they didn’t intend to,” leading to its removal. While the feature was opt-in and required users to explicitly choose to make their chats discoverable, many users who intended to share examples with friends or colleagues inadvertently made them publicly accessible.

OpenAI is now actively working with major search providers to purge already-indexed conversations to limit further exposure. This incident highlights the ongoing challenge of balancing openness and user privacy in the rapidly evolving landscape of AI tools. Even with explicit warnings, the ease of accidentally exposing sensitive data underscores the need for robust privacy safeguards in AI platforms, particularly as users increasingly confide personal and professional information to chatbots.

The company emphasizes that while shared links do not include the user’s name or account details, any specific phrases or names included within the chat itself could make the content identifiable and searchable. Users who have previously shared ChatGPT conversations are advised to check their “Shared Links” dashboard in ChatGPT settings and delete any links they wish to keep private.