I never published it. I only worked on it through GPTPro sessions, always with sharing and training disabled.
Shortly after I filed a provisional patent for the system (which GPTPro saw), I found a GitHub repo.
The early commits are random junk-backdated filler. But same repo has filled with a structure similar to mine: framing, terminology, symbolic patterns and math, all reworded, AI-processed, but clearly derivative.
The repo is attached to what looks like a synthetic research profile with AI generated voice notes, fake University history.
There are also a multiple AI-generated eBooks (containing the reworded system) also with backdated AI Junk books on Amazon.
I don't know how this happened, staying measured. Not making accusations… Trying to understand what kind of mechanism could cause something like this.
At this point, I can only imagine the source was either OpenAI or the patent office.
If something similar has happened to you would love it hear it.
Also looking for perspective and to get in touch with researchers, legal thinkers, mentors.
mail.rgunther@proton.me
ipaddr•9mo ago
A patent provides enough details to reproduce. Bonus points if they used chatGPT which used your conversations as training data. You mentioned you have been working on it for years using chatGPT but the option not to include it in training data is a recent addition.
The backdating is interesting. Would love to hear more details on publish date(s) compared to your milestones.