META To Train Its AI On Europe’s Users Data

The data of the users of Facebook, Instagram will soon be used to train the AI ​​systems of the META group. The company wants to inform its users about this step this week. There is still time to contradict this.

Group contacted users

The announced communication should not only explain which data is used specifically. The company said that a link to a form that can be objected to can object to data processing. Meta assures that both previously submitted and new contradictions are taken into account and respected.

According to the company, the form was deliberately designed to be user -friendly – which is otherwise often avoided for data protection settings. Only public content and interactions with the META -KI – such as postings that are generally visible, or questions about the chat bot are to be used. Private messages and all activities of people under the age of 18 should not be explicitly used to train the systems.

Meta justifies the step with the aim of better gaining its AI models to the needs of European users. “We see it as our responsibility to develop AI that is not only available for Europeans, but was actually made for them,” says one opinion of the company. In order to better understand the cultural and linguistic diversity of Europe, it is necessary to train AI systems with appropriate data – from regional dialects to the country typical of the country to local knowledge.

Broad basis

Especially with a view to multimodal AI models that can process texts, language, images and videos, the broadest possible data basis is crucial. Meta also points out that other companies such as Google and Openai have already used data from European users to further develop their systems.

The current announcement takes place shortly after the publication of Metas new language model Llama 4. This was recently criticized, since an optimized version was apparently used specifically on a valuation platform to achieve better placements.