- Home
- Off Switch Directory
- How to Object to Meta Training Its Models on Your Posts
Off Switch Directory · Messaging, social and creators
How to Object to Meta Training Its Models on Your Posts
In the EU, Meta trains its models on adults' public posts and comments and on chats with Meta AI. You can object at any time with the form in Meta's notice.
- Off switch
- Opt out documented
- Training
- Training opt out in the EU
- Last checked
- Maker
- Meta
Due for a recheck. These steps were last checked on , more than 45 days ago. Menus may have changed.
What it is
The feature, in one line
Meta trains its models on public content shared by adults in the EU and on people's interactions with Meta AI.
Any device
Object to training
- Open the notice Meta sends in the app or by email.
- Use the form linked in the notice to object.
Source: Making AI Work Harder for Europeans Meta, checked
Menu names are shown as the maker's English help page writes them. On a device set to another language they appear in that language, in the same place. The source link opens the maker's page, which offers its own languages.
What changes
What turning it off does, and does not do
In the maker's own terms, from the pages linked below.
What it does
Meta says it honors every objection form it has received and every new one. Private messages with friends and family are not used for training.
What it does not do
The objection form is announced for people in the EU. Public data from accounts of people under 18 in the EU is not used for training.
Where it applies. Meta's announcement covers people in the European Union.
Official sources
- Making AI Work Harder for Europeans Meta · published , checked
The team checks this page each month. The next check is due by . How pages are checked
Questions
Straight answers
Two quick answers.
How do I stop Meta training its models on my posts?
In the EU, Meta trains its models on adults' public posts and comments and on chats with Meta AI. You can object at any time with the form in Meta's notice.
Does Meta train on my private messages?
Meta says it honors every objection form it has received and every new one. Private messages with friends and family are not used for training.
What the directory does
- It shows the controls each maker documents, with the maker's own menu names and a link to its help page.
- It lists features whose maker says they cannot be turned off, or documents only a partial off.
- Every page is checked each month and shows the date. A page older than 45 days says it is due for a recheck.
What it does not promise
- It does not test the apps or prove that a switch works as the maker describes.
- It does not cover every app, plan, device or country, and menus can change between checks.
- Turning a feature off does not delete data a maker already holds, unless the maker says it does.
More switches
Related apps
Other apps of the same kind in the directory.
Meta · WhatsApp
How to Turn Off Meta AI in WhatsApp
No off switch documented
Checked
Google · YouTube
How to Stop Third Parties Training on Your YouTube Videos
Training opt out documented
Checked
Snap · Snapchat
How to Remove My AI From Snapchat
Partial off only
Checked
X · X
How to Stop X Using Your Posts to Train Grok
Partial off onlyTraining opt out documented
Checked