Most AI tools treat whatever you type, upload, or post as training material by default. Opting out is the highest-impact step in Opt Out October, because it stops ongoing collection instead of cleaning up a one-time leak. Note that these companies rename their menus often; if a label here has changed, look for the setting that controls "training" or "model improvement," or follow the EFF's guide.
1. ChatGPT (OpenAI)
Settings > Data Controls, and turn off "Improve the model for everyone." Your future chats stop feeding the model.
2. Google Gemini
Open your Gemini activity settings and pause your activity history. While it's paused, your conversations aren't used to improve the models.
3. LinkedIn
Settings > Data Privacy, and turn off the setting for using your data to improve generative AI. LinkedIn turned this on for most accounts automatically.
4. X (Grok)
Settings > Privacy and safety, and turn off the option that lets your posts train Grok.
5. The ones with no real opt-out
Some platforms, notably Meta AI in the US, don't offer a full opt-out. There, the realistic move is to lock down or remove public content, since public posts are the part they train on. Set your profile to private and prune what's visible.
Do this once, this month
Pick two accounts you use daily and flip their toggles today.
Open the EFF's Opt Out October guide at eff.org for the full list, including Twitch and Snapchat.
Check back when you install a new app, the training switch is usually on by default.
Quick tip
Changing the setting doesn't pull back what a model already learned. That's exactly why turning it off today matters more than waiting, every day you leave it on is more data in.
Up next · Monday
The week's most important privacy story, in plain English: what happened, why it matters, and what to do about it.