AI start-up OpenAI has disabled a short-lived feature in its chatbot ChatGPT that allowed users to make individual conversations discoverable by search engines such as Google.
Announced by OpenAI’s chief information security officer Dane Stuckey on X (formerly Twitter), the change comes after thousands of private chats, some containing personal details, were inadvertently indexed and surfaced in public search results.
Earlier this year OpenAI quietly rolled out an opt‑in sharing option. This feature allowed users to select a chat, tick a checkbox to “share with search engines”, and thereby enable their conversation to appear in Google results.
Although not enabled by default, many users who intended to share helpful examples with friends accidentally made their chats publicly accessible. A Fast Company investigation found over 4,500 ChatGPT conversations already indexed, ranging from innocuous Q&A to discussions containing names, locations and other identifying information.
On a recent podcast OpenAI CEO Sam Altman underscored the sensitivity of ChatGPT dialogues. He said “People talk about the most personal things in their lives to ChatGPT.” That very trust highlighted the hazards of public indexing. Even after a user deleted a shared link, or the chat itself, the content could remain discoverable until search engines recrawled and removed it from their indexes, prolonging the risk of unwanted exposure.
Search Engines to Remove Indexed Chats
Stuckey confirmed that, in addition to disabling the feature, OpenAI is collaborating with major search providers to purge already-indexed conversations. “We think this feature introduced too many opportunities for folks to accidentally share things they didn’t intend to, so we’re removing the option,” he wrote. By completely removing the search‑indexing toggle, OpenAI aims to prevent further inadvertent disclosures.
The incident highlights the challenges of balancing openness and privacy in AI tools. As generative models become integral to personal and professional workflows, even optional features can have wide-ranging consequences. With this feature now retired, OpenAI refocuses on safeguarding user trust while exploring safer ways to surface exemplary public content, without jeopardising the confidentiality that users expect from their AI assistant.
