Overview
- Fast Company identified about 4,500 ChatGPT conversations that were publicly indexed on Google after users checked the discoverability box.
- OpenAI has disabled the “Allow search engines to index this link” checkbox in its share feature following reports of unintended exposure.
- Exposed chats include highly sensitive content such as personal traumas, sexual preferences, corporate secrets, and illegal activity disclosures.
- Existing web caches and archives mean thousands of previously shared links remain accessible despite the feature’s removal.
- Security specialists and ethicists are calling for clearer privacy-by-design defaults and a retroactive cleanup of all indexed conversations.