Since May 27, Meta has been using data from our public posts on Facebook and Instagram by default to train its AI, \u201cMetaAI\u201d.<\/p>\n\n\n\n
This concerns our photos, captions, comments – everything we shared before May 27 and will continue to share publicly after that date, unless we object. In the frantic race for AI, Meta is thus returning to the fray after a suspension of this project in 2024.<\/p>\n\n\n\n
I had the pleasure to speak about this subject on Romain Desarbres’ show.
I reminded him of a fundamental right: we can exercise our right to object at any time, and May 27 is not a cut-off date. I invite you to go to the privacy settings of the two networks.<\/p>\n\n\n\n
In its privacy conditions, Meta asserts its legitimate interest in carrying out this processing. Several questions arise about the legality of this processing under the European Data Protection Regulation (GDPR), including the compatibility of AI training with the original purposes. If this were not the case, this new use of data would be illegal and would require a new legal basis. In this respect, the temporality of posts is eminently problematic. In the past, would we have publicly shared this content if we had been informed that Meta’s AI would be trained on our writings, photos and videos? With regard to the protection of minors, their posts would be protected from this new use, but what about their photos published or republished by adults?<\/p>\n\n\n\n
As Meta’s European headquarters are in Ireland, the Data Protection Commission (DPC) is in the front line to examine these questions.<\/p>\n\n\n\n
While we await the DPC’s conclusions, why are we feeding our European data to the AI of an American tech giant?<\/p>\n\n\n\n
Protecting our data means defending our sovereignty. Let’s support European AI that is ethical and respects our rights.<\/p>\n\n\n\n