A Glitch Caused ChatGPT To Leak Prompts In Unexpected Places, New Reports Claim

This isn't the first time OpenAI has found itself at the center of a story about personal and private chats with the company's AI chatbot somehow leaking to places it shouldn't. But it is perhaps one of the more intriguing reports we have seen, as these claims suggest that ChatGPT chats were actually leaking out into Google Search Console (GSC), allowing people's private queries to the AI to be visible to anyone using the Google system.

For anyone who isn't familiar with GSC, it's essentially a system that allows developers to monitor search traffic. That said, it clearly isn't where you expect your chats with OpenAI's chatbot to end up. Luckily, according to a report from Ars Technica, this issue has since been resolved, and OpenAI says that it was all because of a glitch only affected a "small number of search queries" to be routed through GSC, where they were visible to others.

This particular issue could be tied to OpenAI allegedly scraping Google to answer user prompts. While OpenAI did not confirm or deny the theory, Jason Packer, the owner of the consulting firm Quantable – which was one of the first to outline the issue in October — wrote in the company's report that this behavior could help explain why these queries would show up in GSC.

While OpenAI says it has fixed the issue, Packer doesn't seem convinced it is entirely gone, especially if OpenAI is scraping Google for answers. Online scraping has become an easy way for AI companies like Perplexity to help find the answers to queries presented by its users, and some have even found themselves in hot water over illegally scraping websites that have disallowed the practice.

Like many of the other issues that chatbots like ChatGPT have run into — including calls from the legal system to release full chat histories in some legal cases — this is just another reminder of the ever-changing dangers that surrounding online chatbots and the new part they play in our lives. Some people have come to rely on ChatGPT and other AI chatbots in place of friends or even therapists. And while we've seen OpenAI take harder stances against this, especially after being sued by parents of a deceased teenager, issues continue to crop up.

Using an online chatbot is not private, and you don't really have any control over where your data goes after you put it into the query system. So, if you plan to use ChatGPT or any AI chatbot regularly, be careful what you share with it, because those prompts might just end up somewhere they shouldn't.