OpenAI responds to privacy criticism from international companies and government agencies by addressing two of the biggest criticisms with new data control settings.
First, ChatGPT users can now turn off chat history. Turning it off also ensures that data entered in chat will not be used to optimize future OpenAI AI models.
However, the chat history will no longer appear in the ChatGPT sidebar, so you will lose an important feature. Generated chats with the chat history turned off cannot be restored.
Regardless of the disabled chat history, OpenAI will continue to store the chats for 30 days, but will only evaluate them if there is a suspicion of abuse. After 30 days, the chats will be permanently deleted.
OpenAI will continue to store chats and use them to optimize the model for those who leave chat history enabled.
New ChatGPT data controls
Disabling chat history is available in the ChatGPT settings under Data Controls for both paid and free accounts. Chat history is enabled by default and must be disabled here to prevent your data from being used for continuous AI training.
In addition to disabling chat history, the new data control settings also offer an option to export data and an option to delete your account. Deleting an account will delete all data associated with that account and cannot be undone. This also applies to accounts that use the API. The export function sends all ChatGPT chats and associated data to the account address via email.
More privacy-friendly ChatGPT Business will be available in the coming months
For businesses, OpenAI intends to offer a new service called ChatGPT Business “in the coming months”. By default, user data will not be used to train models, but chat history will be kept. ChatGPT Business for professional users and enterprises uses the same privacy policies as the GPT API.
If you would rather not wait for ChatGPT Business, you can use this form to unsubscribe from model training immediately. After submitting the form, new conversations will not be used for model training.
The new privacy settings are OpenAI’s response to increasing criticism from European data protection authorities, as well as large companies banning their employees from entering corporate data into ChatGPT. The coming weeks will show whether these measures are enough to quell the criticism.