There was an option yesterday, under “advanced options” in “Configure”, for custom GPTs to make sure the data is not used to train/improve OpenAI models.
It’s not there anymore, does anyone know why?
3 Likes
My guess would be that data used in GPT’s may be used for training purposes in this implementation, there may be plans for an enterprise version that allows data to be withheld from training in the future.
2 Likes
Is it just me expecting that a company like OpenAI, in particular for services for paying users, should have pages or docs or something online updating users of the changes they do to beta products, and what they mean?
In particular if they do changes on a service running where users might have already used some data they felt was safe to use with an option that then they take away without explaining what it means?
When developing products that make use of beta services, I think there is an expectation of change inherent in that. While it would be nice to have every change documented, in reality PR’s get ran and options can change without notice. I do agree that such changes should try to be announced and documented, but you should make allowances for the beta nature of the services.
1 Like
I hope at some point they’ll follow up with some explanations, maybe even if cumulative, on how things changed and how they’ll work, because custom GPTs like Assistants are definitely they most important functionalities for the future.
The same happened to me, but I noticed today that option only shows up if you have a file uploaded, or at least that’s the case for me.
1 Like
Good to know it was there at some point. I never saw it. The unhelpful FAQ says it is an option:
If I build a GPT, can I opt out of training?
OpenAI has introduced a GPT-level opt-out option for builders. This feature allows builders to decide whether their proprietary data can be used by OpenAI for model training.
It would be great to know more about this.
Hello! Are there any updates about this? I created a custom GPT yesterday and had the option to opt out of having my data used for the model’s training. Today, I noticed that the option is no longer available, neither for my built GPT nor for new GPTs.
Does anybody know if pre-existing GPTs where the advanced option to opt-out of training was selected will continue to be opted-out, or will they now be fair game for training data?
Does anyone have any additional information to share on this topic? Thanks in advance!
I guess you have to get in touch with privacy@openai.com to get clarification on this matter. You could also request your data and check for yourself. Quite frustrating.
If anyone hears from OpenAI on this, please report back here.