100,000 ChatGPT Chats Exposed: Why a New Feature Backfired on OpenAI

100,000 ChatGPT Chats Exposed: Why a New Feature Backfired on OpenAI

OpenAI’s testing of a feature allowing search engine indexing of user chats led to over 100,000 conversation records becoming publicly accessible. Although the feature was disabled and efforts were made to de-index these links, many remain accessible on platforms like the Internet Archive and Bing. #OpenAI #ChatGPT #privacybreach #informationleak

Keypoints

  • OpenAI tested a feature that permitted search engines to index shared ChatGPT conversations.
  • Over 100,000 chat records became publicly searchable before the feature was disabled.
  • Many users unknowingly enabled search engine crawling despite warnings, due to a misjudgment of user understanding.
  • Some indexed conversations contained sensitive information, including details of a legal case involving indigenous land exploitation.
  • Efforts to remove indexed links were limited to Google, leaving others like Bing and the Internet Archive with indexed content.

Read More: https://securityonline.info/100000-chatgpt-chats-exposed-why-a-new-feature-backfired-on-openai/