Recently, Meta announced that it wants to use posts and images from users to train AI models. What those AI models are supposed to do, and what your Facebook and Instagram posts have to do with it, remains rather vague. That's why we urge you to object to the use of your posts; we'll explain how.

Meta's new AI

Meta develops generative AI. At Facebook's privacy center, Meta explains that effective generative AI models must be trained with big data. The bottom line: Meta wants to use all the(public) photos and posts that you have ever posted on Facebook or Instagram to train AI.

And for what purpose? That remains rather vague. Meta claims that using your data is in the legitimate interest of Meta, their users and other people. But what that legitimate interest is also remains unclear. What exactly Meta intends to do with that generative AI next is a mystery.

All in all, Meta is going to train "something" by using your photos, and doesn't explain what it is. Translating this to the offline world: a very rich person walks into your house, grabs your photo books and leaves. In the doorway he shouts "I'm using these for something cool I want to develop." It seems clear to us: you don't want this.

What to do?

Because Meta claims to have a legitimate interest (in our opinion, this is nonsense), the use of your data is regulated via an opt-out. Your data will be used for Meta's generative AI unless you object. As far as we are concerned, it would have been better if that had been the other way around, an opt-in: your data is not used for Meta's generative AI unless you indicate that you want it to be.
However, right now, the responsibility to ensure that your century-old Facebook posts are not used for Meta's latest project lies with you.

Surely they won't get away with this?

Obviously, it is absurd that Meta wants to introduce AI in this way. Especially since Meta claims to have a legitimate interest in processing this data. Fortunately, we are not the only ones who feel this way; None Of Your Business (NOYB) filed complaints with 11 privacy watchdogs in the EU about this practice. And with results: training of Meta's generative AI models has been delayed because the Irish privacy watchdog has put them on notice.