OpenAI removes ChatGPT feature exposing private chats on Google Search
What's the story
OpenAI has officially announced the removal of a controversial feature that indexed ChatGPT conversations on Google Search, Bing, and DuckDuckGo. Dane Stuckey, OpenAI's Chief Information Security Officer (CISO), took to X to announce the decision. The update, which was quietly introduced earlier this year, sparked major privacy concerns after several private chats were found in search results. The feature also created a link for sharing on platforms like WhatsApp, Instagram, or Facebook.
Feature explanation
Stuckey explains reason behind removal
Stuckey explained that the feature was an experiment aimed at helping users discover useful conversations. It was an opt-in "Make this chat discoverable" option where users could share their chats with search engines by selecting a specific chat. However, he admitted that it created too many chances for unintentional sharing, as users accidentally checked this box, thinking it was necessary for visibility. Stuckey also revealed that the company is working with search engines to get the already indexed content removed.
User concerns
Potentially harmful conversations indexed on Google
OpenAI CEO Sam Altman noted in a recent podcast that "people talk about the most personal shit in their lives to ChatGPT." A Fast Company report revealed over 4,500 indexed conversations on Google. While some were innocuous, others contained identifiable information such as names and places that could potentially compromise user safety. Even deleting a conversation or its link wouldn't guarantee its removal from Google's index.