Facebook, Instagram & Co: Soon food for AI?

From May 27, 2025, the time has come: Meta, the company behind Facebook and Instagram, wants to start using posts from European users to train its artificial intelligence. This is not about private chats - but about publicly posted content, i.e. everything that was or is visible to others on social networks.

The catch: anyone who does not want this must actively object. Meta relies on a "legitimate interest" - a term that exists in data protection law, but is a constant source of debate. The Consumer Association of North Rhine-Westphalia takes a critical view of this - and has now applied to the Cologne Higher Regional Court for a temporary injunction. This is intended to stop the use of data in the short term before facts are created.

Data protectionists: "Once in the system - never out again"

According to the consumer advocates, the big problem is that once a data set has been fed into an AI model, it is no longer possible to undo what the machine has learned from it. Christine Steffen from the consumer association also emphasizes: "With the application for a temporary injunction, we want to prevent Meta from creating facts before the legal situation has been clarified."

The aim is not to block the development of artificial intelligence - but to put it on a fair and lawful footing. The central demand: users should retain full control over their data - and not have to realize afterwards that their posts have long been circulating in an AI world over which they no longer have any influence.

Meta: "Everything by the book"

Meta itself rejects the accusations. The company explains that it complies with the requirements of the European Data Protection Committee and works closely with the Irish Data Protection Commission - which is responsible for Meta in the EU. In addition, the planned use was dealt with openly and users were informed via in-app messages and emails. Anyone who does not want to can object.

And: AI training with public content is the industry standard, says Meta. This is the only way the technology can be further developed in a culturally and linguistically appropriate way for Europe. The company argues with progress and competitiveness - but that is not enough for consumer advocates.

What does this mean for users - and what can be done?

The good news: it's not too late! If you don't want Meta to use your own public posts to train its AI, you can object - until May 27. The NRW consumer advice center offers simple instructions on how to object on its website.

What many people are not aware of is that even seemingly harmless contributions - such as a public comment or a picture with a caption - can be incorporated into AI systems and have a lasting influence on their behavior. It is therefore important to make a conscious decision as to whether you want to be part of this development.

AI must not be a self-service store for data giants.

Let's be honest: a company with the power and size of Meta should not hope that users will be too comfortable to object. If you really want to act transparently, ask first - and then act, not the other way around.

Just because something is technically possible and economically attractive doesn't mean it's legally or ethically okay. The controversy surrounding meta-AI shows this: The debate about data protection is not over - it is just beginning.