ChatGPT bug temporarily exposes AI chat histories to other users
The chat history sidebar is currently unavailable while OpenAI fixes the feature. The bug reportedly showed users the titles of others’ conversations, but not their contents.
Illustration by Alex Castro / The Verge
ChatGPT’s chat history feature is currently offline as of Tuesday morning after a bug exposed brief descriptions of other users’ conversations to people on the service.
On Reddit, one user posted a photo showing descriptions of several ChatGPT conversations they said weren’t their own, while someone else on Twitter posted a screenshot of the same bug. A spokesperson for OpenAI confirmed the incident to Bloomberg, noting that the bug did not share full transcripts of conversations but only brief descriptive titles.
The bug is an important reminder to be careful about the sensitive information shared with ChatGPT. “Please don’t share any sensitive information in your conversations,” warns an FAQ on OpenAI’s website, which notes that the company cannot delete specific prompts from a person’s history, and that conversations may be used for training. However, there will inevitably be a strong temptation to share confidential information with the chatbot, particularly as businesses continue to experiment with how to make use of the new tool.
Bloomberg reports that OpenAI temporarily shut down ChatGPT on Monday in response to the bug, but that it was brought back online later that night. As of this writing the chat history sidebar has been replaced with a message noting that “History is temporarily unavailable” and that the company is “working to restore this feature as soon as possible.” The last update on OpenAI’s status page from 10:54PM ET on Monday notes that service has been restored, but it’s still working to bring back past conversation histories for all users.
The cause of the issue is thought to be a bug in an unnamed piece of open-source software, Bloomberg notes, though an investigation into the precise cause is ongoing.