Meta Plans to Use EU User Data for AI Training

Recently, Meta announced its plan to use the data of EU users from its apps, including Facebook and Instagram, to train artificial intelligence (AI) systems. The data covers public posts, comments, and chat records with Meta AI, but excludes private messages exchanged with friends and family. The company emphasized that the data will only come from users who are at least 18 years old.

Starting this week, Meta plans to inform EU users about the initiative through in-app notifications and emails, providing a link for them to opt out. According to its privacy policy, users can submit their objections online. Currently, the company remains in a delayed implementation phase due to regulatory feedback, and the exact timeline for execution has not been determined. Last year, Meta suspended its AI training efforts in Europe following requests from Ireland's regulatory authority.

Meta stated that utilizing EU user data to train AI aims to enhance the model’s regional adaptability, including dialects, slang, localized knowledge, and humor expressions specific to different countries. This move is particularly important for multimodal AI systems that integrate text, voice, video, and images.

Previously, Meta announced a similar training initiative using UK user data. Both the EU and the UK have stricter personal data protection regulations compared to the US. Notably, Meta acknowledged last year that it had already used all publicly available text and photos posted by adult Facebook users since 2007 to train AI models, meaning the scale of newly added data may be relatively limited this time.