Meta to use public data of its users in the EU to train its AI models

(AFP) Meta Platforms, the parent company of Facebook and Instagram, will train its generative artificial intelligence (AI) models from European users' public content and their conversations with the Meta AI chatbot, the group announced Monday.

European users of the group's platforms (Facebook, Instagram, Messenger and WhatsApp) will receive a notification in the coming days informing them of this new policy, which will also include a link to a form to refuse to participate.

"EU-based individuals using our platforms can choose to object to their data being used for training purposes," the company added.

Specifically, Meta AI will be able to train from public user posts and comments (such as captions on Instagram images) and from requests and questions sent to Meta AI, the conversational bot available in the EU since the end of March.

The change should not concern children under 18 or private messages exchanged by adult users "with their friends and family," the social media giant said.

The launch of Meta AI in the EU was delayed for more than a year because of European regulations on new technologies, such as the General Data Protection Regulation (GDPR) and those on digital marketplaces and AI.

Meta relies on a decision by the European Data Protection Committee, which in December found that "legitimate interest" could provide a sufficient legal basis for developing and deploying AI models in the EU.