Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I found that ChatGPT was trained on private Google Drive data. I prompted it to help me create a murder mystery and became suspicious from its answers. I then copied a short setting description from one of my mysteries which I stored on Google Drive and asked it to generate smaller storylines. It described the storylines I had written exactly.


Why would Google share data with a direct competitor to help them train their model?


I have no idea. However, the storylines were not obvious, and yet they were exactly the same. The probability of this is very low.


Sorry, but that's highly unlikely. A better response is to reflect on the origin and the amount of un-obvious-ness of those storylines.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: