As a dev who personally has been using ChatGPT for helping with code, this sounds pretty unrealistic.

GPT struggles with pretty basic stuff still, and pretty notoriously produces incorrect/unsafe/invalid code and constantly hallucinate wrong stuff into its output.

I’m having a very hard time believing it produced an entire functional app that didn’t constantly require intervention by a skilled dev to repeatedly “course correct” it back to actually working.

What scares the shit out of me is people doing shit like this to try and make apps that are going to actually collect people’s important info, like credit card #s, logins, pictures of their ID, etc etc.

ChatGPT has yet to produce anything remotely coherent when it comes to any actual security. And I use C# + JS for my projects, not exactly esoteric topics and should be areas it has heavy training on.