A bug exposed the titles of some users’ past conversations with OpenAI’s ChatGPT to other people using the AI chatbot, according to a report from Bloomberg. OpenAI temporarily disabled ChatGPT on Monday while it fixed the issue.
When someone is using ChatGPT, a chat history with the service shows up along the left-hand side of the website, letting people continue past conversations. OpenAI confirmed to Bloomberg that some people saw the titles of other people’s chats with ChatGPT instead of their own. However, the “substance of the other users’ conversations was not visible,” according to Bloomberg.
In a tweet on Wednesday, OpenAI CEO Sam Altman said the “significant issue in ChatGPT” was the result of a “bug in an open source library.” He confirmed that a “small percentage” of users were able to see other users’ conversation history.
While it’s unclear exactly how many people were impacted, it’s a reminder to be cautious when using ChatGPT and other new AI tools, many of which are still in beta or testing phases. In an FAQ about ChatGPT, OpenAI says it’s unable to delete specific prompts from a user’s history, adding that people shouldn’t “share any sensitive information in your conversations.”
OpenAI released ChatGPT late last year, and by January, the service was estimated to have reached 100 million active users. It’s been followed by a surge of new AI tools and services from companies including Microsoft, Google and Adobe.
ChatGPT was back online Monday evening, but OpenAI said on a status page that it was still working to fully restore conversation history to users.
Editors’ note: CNET is using an AI engine to create some personal finance explainers that are edited and fact-checked by our editors. For more, see this post.
