OpenAI’s license forbids the creation of competing models with output from OpenAI’s models. Does this apply to OpenAI’s fine tuned models? I’m building an AutoGPT like AI. The objective is to solve complex problems that can not be solved with a single query to the LLM by making many queries to the LLM. If I’m using GPT-4o as the LLM by my AI, can I use records of my AI’s activity to fine-tune a GPT-4o? I would think that OpenAI would allow this, they are still going to be serving the fine-tuned model so users will still be paying them. It would be great if someone from OpenAI could clarify this legal point.
This question is not just limited to an individual user’s concerns; it might reflect the questions and interests of other users as well.
So, if it’s just a preliminary view, it would be helpful for OpenAI to provide a response.
1 Like