Meta has announced that it will begin training its AI models using public content from European Facebook, Instagram, and Threads users. This includes posts, comments, and captions, as well as the history of conversations with its chatbot. However, it will not affect private messages or the content of minors' accounts.
This decision comes after confirmation from a panel of EU privacy regulators, highlighting Meta's approach to meeting legal responsibilities. The move has sparked controversy, but it is also a crucial step forward in Meta's plans to consolidate its position in the competitive world of AI.
The company maintains that its goal is to make its AI more representative of the different cultural, linguistic, and social nuances of the areas where it operates. In a statement, Meta said: “We believe we have a responsibility to develop AI that is not only available to Europeans, but built for them.”
Training models with public data is not exclusive to Meta. Companies like Google and Open AI have also used user-generated content to enrich their AI models. However, Meta has emphasized that its method is more transparent and that users cannot be involved in this process.
Organizations like the European Center for Digital Rights (NOYB) have filed complaints in 11 EU countries. The company claims that Meta deliberately hinders the opt-out process through “dark patterns,” design strategies that weaken privacy protection, such as difficult-to-find forms or the requirement to provide a reason for opting out, even if any justification is accepted.
Meta will start to inform those EU users via emails and notifications on their platforms. These will include a link to a form to opt out of using your content in AI training. In addition, this form will be available through their privacy policy, where modifications have already begun to be reflected. Meta has indicated that it will honor all opt-out requests.
The procedure for objecting is:
It is important to act before Meta's deadline, as after that time, the company states that it will not be able to guarantee the deletion of the data previously used in training its AI models.
Although Meta claims that this training will facilitate AI that is more in tune with European diversity, many users are concerned about the consequences in terms of privacy. The ruling opens up a new front in the discussion about how big tech uses personal data and to what extent users can have absolute control over it.
The regulatory context, stricter than in other areas, has forced Meta to delay these plans. Now, after obtaining the approval of the European Data Protection Committee, the company is ready to move forward. However, the focus will remain on how it implements this change and whether it truly ensures users' right to decide.
