ChatGPT bug leaks private conversations of users, OpenAI issues warning
OpenAI, the artificial intelligence research company behind the wildly popular AI chatbot service, ChatGPT, announced its first widely known privacy glitch on Wednesday. In a tweet, Sam Altman, chief executive of OpenAI, confirmed that the service saw some private chat information of users being leaked online as a result of a bug in an open source library. While no critical private information has been leaked online, the issue shows how privacy issues can be critical in public AI chat applications such as ChatGPT.
What the ChatGPT privacy issue was
The ChatGPT privacy bug surfaced earlier this week, when users on Reddit and Twitter started sharing screenshots of chat titles that other users had on the app. To be sure, chat titles appeared in the chat history section of ChatGPT, listing down the main question of every chat session that a user had had with the application.
In response to this issue, a report by Bloomberg said on Monday, March 20th that OpenAI decided to shut down ChatGPT, before bringing the service back online later in the day. However, the chat history section has since been disabled, and presently remains out of access for users. It’s not clear if the same holds true for paying users with access to ChatGPT Plus as well.
In a statement on the issue on Wednesday, March 22nd, Altman said on Twitter, “We had a significant issue in ChatGPT due to a bug in an open source library, for which a fix has now been released and we have just finished validating. A small percentage of users were able to see the titles of other users’ conversation history. We feel awful about this.” He further added in a separate tweet that users will not have access to their chat histories for approximately nine hours from March 20th, likely owing to a restoring activity that may bring chat history back to the platform.
“We will come back with a technical postmortem,” he added.
The issue is the first notable outright privacy issue faced by ChatGPT, but was not a hack or an orchestrated cyber incident. ChatGPT has so far been flagged by users in terms of AI bias and safety of responses — factors that the company has said will be resolved by the latest generation of its underlying language model, GPT-4.
The post ChatGPT bug leaks private conversations of users, OpenAI issues warning first appeared on 91mobiles.com.
https://ift.tt/ax2ZUKE
https://ift.tt/LBu0KIa
No comments