OpenAI removed a feature allowing private ChatGPT chats to appear in Google searches after backlash over privacy concerns.

The revelation that ChatGPT conversations were unexpectedly appearing in Google search results has prompted OpenAI to swiftly retract a contentious feature. This development emerged after some users inadvertently allowed their private and personal exchanges to become publicly searchable.

A detailed investigation by Fast Company uncovered the privacy breach, highlighting how thousands of ChatGPT conversations made their way into Google's index-potentially just a fraction of those "visible to millions." Although these indexed chats did not contain identifiable user data, they often included intimate details such as descriptions of relationships with friends and family, which could inadvertently lead to identification.

Dane Stuckey, OpenAI's chief information security officer, clarified on X that affected users had opted in for indexing by selecting an option when choosing to share a chat.

The report also noted a common practice among users: sharing chats on platforms like WhatsApp or saving links for later reference. However, Fast Company observed that users might have been confused by the wording used:

"When users clicked 'Share,' they were presented with an option to tick a box labeled 'Make this chat discoverable.' Beneath that, in smaller, lighter text, was a caveat explaining that the chat could then appear in search engine results."

Credit: ChatGPT Share box via Dane Stuckey on X

Initially standing firm on their stance that the labeling was "sufficiently clear," OpenAI later acknowledged through Stuckey that "ultimately," this functionality opened up too many risks of unintended sharing. The mishap involved sensitive topics like drug use, sexual activities, mental health issues, and traumatic experiences.

Carissa Veliz, an AI ethicist at Oxford University expressed her surprise about Google's indexing of these highly sensitive conversations.