Meta (Facebook / Instagram)
Meta's captured privacy policy says it collects what you exchange with "AI at Meta" but never once says "train". What happens to those messages afterwards isn't stated in it.
This clause is real, but it doesn't resolve the training question — see the finding below.
"Interactions with AI at Meta and related metadata. For example, information you or others exchange with AI at Meta like content and messages."
- 01 Jun 2026
- Facebook & Instagram (consumer accounts)
- global
- FA5E1B2A…5DB1
Why this is still UNCLEAR: The document names AI repeatedly as a product surface and is silent on it as a training input. It lists "Interactions with AI at Meta and related metadata" among the information collected, and elsewhere describes using information to "Understand and enable creation of content like text, audio, images and videos, including through AI at Meta" — but no form of the word "train" appears anywhere in the capture, and neither does "model". The single mention of artificial intelligence outside the product name is a line about supporting research. A permission to train models on user content cannot be read into a description of what is collected, so this stays unresolved rather than being graded on inference. Note also that this quotes the June capture, not the two most recent ones: those were served to the monitor in Russian (see the note on this verdict's freshness).
What each plan can refuse
The same questions asked of every plan we could identify. Every cell is sourced from a quoted clause or a documented search — see the evidence above.
- Not established
- Not stated
- —
- Not established
Everything we read
Vendor response
Meta (Facebook / Instagram) has not disputed, clarified, or announced a change to any clause quoted above. If that changes, it will be published here verbatim.
Report a wrong clause