OpenAI: Sorry, ChatGPT Bug Leaked Payment Info to Other Users

OpenAI is confirming that a glitch on Monday caused ChatGPT to also expose payment details for paid users, in addition to leaking conversation histories from random users. 

On Monday, users who tried to subscribe to the paid ChatGPT Plus service reported seeing email addresses from random users pop up in the payment form. But it turns out ChatGPT exposed even more info from paid users. 

After initially confirming the conversation history leak, OpenAI published a more in-depth blog post(Opens in a new window) today going over Monday’s outage, which involved a software bug that caused ChatGPT to leak information on its internal database.  

“Upon deeper investigation, we also discovered that the same bug may have caused the unintentional visibility of payment-related information of 1.2% of the ChatGPT Plus subscribers who were active during a specific nine-hour window,” the company said. 

“In the hours before we took ChatGPT offline on Monday, it was possible for some users to see another active user’s first and last name, email address, payment address, the last four digits (only) of a credit card number, and credit card expiration date,” OpenAI added. “Full credit card numbers were not exposed at any time.”

Tweet(Opens in a new window)

However, the company says the chances of a stranger actually viewing all this payment info from a random subscriber is “extremely low.” That’s because the exposed payment details partly arrived through confirmation emails for new ChatGPT Plus subscribers sent on Monday morning, between 1 a.m. and 10 a.m. PST. 

“Due to the bug, some subscription confirmation emails generated during that window were sent to the wrong users,” OpenAI said. “These emails contained the last four digits of another user’s credit card number, but full credit card numbers did not appear. It’s possible that a small number of subscription confirmation emails might have been incorrectly addressed prior to March 20, although we have not confirmed any instances of this.”

Other exposed payment details were available if a user clicked on the “My account” function on ChatGPT’s website, and then “Manage my subscription” between the 1 a.m. and 10 a.m timeframe. “During this window, another active ChatGPT Plus user’s first and last name, email address, payment address, the last four digits (only) of a credit card number, and credit card expiration date might have been visible,” the company said. 

In response, OpenAI is reaching out to affected users about the potential data breach. “We apologize again to our users and to the entire ChatGPT community and will work diligently to rebuild trust,” the company wrote. 

OpenAI is blaming the leak on a bug in an open-source library to run a database from Redis. The company has been relying on a Redis library to cache user information on its servers. The library is designed to treat both requests and responses as “two queues.” But a problem can occur if a request is canceled before it’s fully processed. 

“If a request is canceled after the request is pushed onto the incoming queue, but before the response popped from the outgoing queue, we see our bug: the connection thus becomes corrupted and the next response that’s dequeued for an unrelated request can receive data left behind in the connection,” OpenAI says. 

On Monday at 1 a.m., OpenAI said it introduced a server change that caused a spike in Redis request cancellations, which caused the data corruption. “This created a small probability for each connection to return bad data,” it said. 

The company has patched the bug and added safeguards to ensure requests to the Redis caches matches the requesting user. “We are confident that there is no ongoing risk to users’ data,” OpenAI said. In addition, the chat history sidebar appears to be restored on ChatGPT.