I’m trying to figure out if files used to fine-tune a model or create an assistant are used to train the model. I know chat history is not but I’m trying to understand the implications of using files that may contain potentially sensitive information (in this case product pricing and details) to help our team with extracting information from quotes and answering technical questions about our products. Apologies if I’ve missed something and this question is answered plainly elsewhere 
Even if the files are placed nowhere else except a temporary data store and then deleted, when the AI receives data from files and responds about them, that language data will be data retained for up to 30 days by OpenAI for policy enforcement and other internal requirements. API language data will not be used to train AI models.
1 Like