What Happens to My ChatGPT Data?

One question I get a lot from educators on using ChatGPT and GPT-powered tools is: will my data be used in training future models?

‍

Data privacy and ownership is probably the second hottest topic after GenAI these days, and it is very important that educators understand OpenAI’s stance on data retention and usage.

‍

I use ChatGPT on a personal account; what happens to my data?

‍

When it comes to ChatGPT, whether you are on the free or the plus version, ‘Chat history & training’ is on by default. Any data you pass over to OpenAI (prompts, context, etc) can be used to train future models. If you would rather keep your data private so that OpenAI won’t use it to train future models, you can turn that off by going to your profile > Data controls > toggle off ‘Chat history & training.’

‍

‍

I will take TimelyGrader as an example. Most, if not all, third-party tools utilize a model via an API integration. When TG passes data to OpenAI for analysis and when they return a response to us, that data is encrypted and will not be used in the training of future models. OpenAI will store that data for 30 days and then promptly delete it. You can read the full policy here.

‍

You can understand why this is so paramount for OpenAI to do with many government agencies, hospitals, and other sensitive institutions using ChatGPT or API integration.

‍

You can rest assured that whatever assignment information, grading rubric, and student submissions you upload to TG will not be used in the training of new models.

‍

Chris Du

CEO, TimelyGrader.ai