Improve model for everyone
What ChatGPT's Improve model for everyone setting actually does, why privacy still matters if you think you have nothing to hide, and how to disable it.
A bit of a rant first, and why not.
Yes, like you, I've been pwned on the internet countless times. But that doesn't mean I (and you) should be less careful with our data. Some people say, "I've got nothing to hide, so what if someone has my data or knows what I'm searching or asking GPT to write for me? Or if GPT is being trained on my prompts and interactions?"
It's not about whether you've got anything to hide. By default, you should prioritise privacy, how organisations use your data, and the security of your information.
Now to the topic at hand. Everyone loves the lightning-fast performance of GPT-4o, the latest omni model. But there is still very little discussion about ChatGPT data controls. Many users I know or have spoken with are not aware of the privacy controls available, particularly the Improve model for everyone option.
The full Medium version explains exactly where the control lives, what enabling it means for training use of your chats, and the steps to turn it off, with the author's notes and images kept on the original post.