It happened to me yesterday and it happened to a friend of mine today.<p>We uploaded a file, we asked a question and we got replies that are clearly not meant for us. My friend even got a document that was clearly uploaded by a lawyer from our city (probably by coincidence).<p>I got replies in English because my message was written in English, and my friend got replies in Italian because his message was written in Italian. We tried multiple message. I did manage to get one correct reply, but meanwhile I got multiple replies that were clearly not meant for me.<p>I would like to know if someone else is experiencing this and where can I report this bug because I bet OpenAI doesn't want this to happen.<p>I'm based in Italy btw.
If you have screenshots or other proof I'd try to contact their security team directly. <a href="https://openai.com/.well-known/security.txt" rel="nofollow">https://openai.com/.well-known/security.txt</a> But might also end up with auto-reply emails to use their bug tracking system
I had an issue where one completely unrelated conversation about a topic bled into one that happened days ago. When I asked about it, ChatGPT said:<p>> It looks like I accidentally analyzed and described a Hammerspoon script instead of pulling data from your science benchmark PDFs—sorry about that!<p>Very weird.
I remember seeing something similar happen around the time OpenAI launched ChatGPT. I can't seem to find it right now, but at one point they served users' entire chat histories to the wrong users, even in different languages.