With the use of ChatGPT through the roof, OpenAI has decided make a long-awaited update focused on privacy. Starting today, users have control over how their conversations are used with the AI chatbot.
Specifically, it is possible to disable saving of chat history with ChatGPT. This not only prevents the information from being stored indefinitely on the OpenAI servers, but it also prevents it from continuing to be used to train the platform.
Anyway, it is important to mention that the ad has a trick. Although Sam Altman’s firm ensures that he will not use the conversations to train ChatGPT, he will continue to have temporary access to them even if the history is disabled.
According to the startup, all new conversations generated with the chatbot after disabling history will be kept for 30 days before being permanently deleted. In this way, OpenAI ensures that it will review them only to monitor possible cases of abuse.
ChatGPT’s new privacy feature is far from perfect, but it’s better than nothing. It is clear, in any case, that even if the ChatGPT history is disabled, you should not share sensitive information —both personal and work— with the chatbot.
Another novelty that comes to ChatGPT is a button for users to export your data. It will make it easier to get a copy of everything that OpenAI stores and uses to train its language models. When requesting the export, the interested parties will receive by email a file with all their conversations and other “relevant information”.
How to disable ChatGPT history
Disabling ChatGPT history is really easy. We just have to go to the lower left corner of the screen, touch on our name and access Settings (Settings). Once there, we will see an option called Data Controls (Data controls) and click on Show (Show).
There a button will be revealed that will allow us disable history and prevent our new conversations from being used to train ChatGPT. In the event that we want to export the chats and other data linked to our account, we will find the Export data option within the same box.
It’s good that OpenAI is starting to provide more tools for users to control how much they share with ChatGPT. As we said above, it’s far from a perfect solution; however, it can help people feel more secure when interacting with the chatbot. Especially after a recent and highly controversial AI glitch.
The incorporation of this new option could also be part of a strategy to avoid further regulatory scrutiny. Let’s not forget that ChatGPT was temporarily blocked in Italy for unauthorized collection of personal information, later used to train artificial intelligence. While other countries have not been so drastic, but have expressed concern about it.
OpenAI works on a new subscription for professionals
In addition to the privacy news, OpenAI announced that it is working on a new subscription called ChatGPT Business. It will target “professionals who need more control over their data, as well as companies looking to manage that of their end users.”
According to the American company, this new subscription will obey the data use policies of its API. This means that end-user information that companies provide to the chatbot will not be used to train it. However, ChatGPT Business will only be launched in the next few months.