A bug in ChatGPT exposed what you wrote to foreign users

by time news

The history feature is turned off. Screenshot: Gigtime

Say congratulations to OpenAI after its first privacy bug was discovered today (Tuesday). As a result, part of the ChatGPT user interface was disabled.

It all started when users on Reddit and Twitter started posting screenshots showing that they had been exposed to strangers’ prompts on ChatGPT. The foreign chats appeared to users in the conversation history feature, which is located on the left side of the interface, including all the searches, prompts and requests made by the users.

As a result, OpenAI disabled the service completely in the morning hours (Israel time) and then neutralized the feature. As of this moment, you still cannot access your search history and instead a message is displayed: “History is temporarily unavailable”.

Screenshot: Jordan L Wheeler

Although the strange bug revealed only the search title and the prompt that the users entered – but not the answer; But once you have the prompt, you can enter it yourself in the tool and get a similar answer yourself – even if not the same. An OpenAI spokesperson confirmed the incident to Bloomberg.

You may also like

Leave a Comment