A serious vulnerability exists in ChatGPT: users can see other people's chat titles

ChatGPT’s historical chat feature has been offline since Tuesday morning after a glitch allowed people to see short descriptions of other users’ chats with the AI.

A user posted a photo on Reddit. He said that a description of the chat content that was not his own appeared in the ChatGPT chat history column, and someone also posted it on Twitter. Screenshot of the same glitch. An OpenAI spokesperson confirmed the incident to Bloomberg, noting that the glitch did not show a full transcript of the conversation, just a simple caption.

This glitch is an important wake-up call for users to be careful not to share sensitive information on ChatGPT. "Please do not share any sensitive information in your conversations," a FAQ on the OpenAI website also warns, noting that the company cannot delete specific content from someone's history and that conversations may be used for training .

OpenAI temporarily shut down ChatGPT on Monday as it responded to the glitch, but reopened it later that night, Bloomberg reported. As of the time IT House wrote this article, the historical chat record sidebar has been replaced with a message, "History records are temporarily unavailable, and we are working hard to restore this feature as soon as possible.".

It is speculated that the cause of the problem is a bug in an unnamed open source software, although the company is still conducting a detailed investigation into the cause.

The above is the detailed content of A serious vulnerability exists in ChatGPT: users can see other people's chat titles. For more information, please follow other related articles on the PHP Chinese website!