But we are unsure if we want to proceed. Is chatGPT grabbing our data from the drives we give it access to and then storing it in their infrastructure too, or are they just running it through embeddings, storing that somewhere and when there are hits on it, use the metadata returned from that to pull in the documents/data directly from the drive during point of use?
If they are just storing our data on their servers, that seems like a pretty steep ask as far as trust. I have searched but not found any relevant documentation that answers this question.
The first thing I noticed is that this feature will only be available to Teams and Enterprise users (in the future). Therefore, any data or information you provide should adhere to the existing data privacy agreement applicable to your entire Teams account.
In my view, this involves the same level of trust as when you send individual prompts or knowledge files to OpenAI, either through chat or when creating a custom GPT.
The difference is likely the scale at which you make your internal knowledgebase available to OpenAI but the underlying trust mechanism is essentially the same.
I hope this helps!
Ps. You are likely already aware of this link, which should help you assess the situation.
Thanks. There is just something about having all of our Google Docs now sitting on another server that bothers me. Thanks for the link. I have signed up. Maybe some of their data diagrams will answer this.
No problem. From a process perspective, it makes sense to me because cloud storage, like Google Drive, is not the most efficient database solution to pull different bits of information from your files.
I also believe that the usual 30-day retention period before auto-deletion applies to the synchronization feature. But this is something I would preferably confirm with a product specialist.