In a series of bizarre incidents, several ChatGPT users reported being able to read strangers’ conversations on their chat windows. These incidents apparently took place on Monday even as OpenAI’s ChatGPT experienced an outage for a few hours.
The anomaly came to light when a ChatGPT user took to Reddit to check whether others were experiencing similar discrepancies in their conversation with the AI-powered bot. As per screenshots shared by the user, the left-side panel which shows the user’s conversation history with ChatGPT included conversations that were not made by the user. The most popular reasoning behind this is that the foreign conversations were made by other users and were then subsequently leaked because of privacy issues with ChatGPT.
Some users shared screenshots on Twitter as well, to show that they were witnessing conversation history in foreign languages that they never use. Since OpenAI has not commented on this issue, it is not yet clear whether there was indeed a data breach or if ChatGPT was making up conversations on its own.
Interestingly, during the ChatGPT outage that took place on Monday, some users had reported that their past conversations with the bot went missing from their chat windows. It is now a matter of speculation whether the conversations that went missing during the outage resurfaced on other users’ screens.
If it is indeed a case of data breach, then users of the widely popular chat bot have several reasons to be worried about. In a bid to make the bot’s answers most customisable to the users' requirements, many users share a variety of personal information in their conversations. A data breach would compromise such critical information as well.
Going ahead, as the AI chatbot war heats up between Google, Micrsoft and others, it will be important to see how ChatGPT responds to these allegations of data breach.
