New ways to manage your data in ChatGPT
OpenAI
OpenAI now lets you switch off chat history in ChatGPT. That means you get to pick which chats help train future models.
Based on reporting by OpenAI — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
OpenAI just handed ChatGPT users a lever they've been asking for: a toggle to turn off chat history. Flip it, and that conversation won't be used to train future versions of the model. Simple as that.
For a company whose entire business rests on hoovering up data to make its systems smarter, this is a notable concession. Every exchange users have had with ChatGPT has, by default, been fair game for training. Now there's an opt-out, conversation by conversation, rather than an all-or-nothing setting buried in account preferences.
The timing isn't random. Regulators in Europe and beyond have spent the past year pressing AI companies on exactly this question: what happens to the things people type into a chatbot, and who gets to decide. Giving users granular control over their own chat history is as much a policy move as a product one.
It also nudges ChatGPT a little closer to how people already think about messaging apps, where you expect some control over what's kept and what's deleted. Whether most users will actually bother flipping the switch is another matter entirely. Defaults tend to win, and OpenAI knows that better than anyone.
My take — AI-written commentary, not fact-checked reporting
I'll believe this matters once the default flips to off, not on — until then it's a nice option for the three percent of users who read privacy settings. Companies love shipping controls that sound like empowerment but really just shift the burden of consent onto people too busy to use them, and OpenAI isn't breaking that pattern here.
Read more about this at: OpenAI