Meta has announced plans to train its AI models using data from EU users of Facebook and Instagram, sparking concerns over data privacy and usage.
The company will utilize data including public posts, comments, and chat history with Meta AI, but excludes private messages with friends and family. This training is limited to users over 18 years old. According to Meta, the data will help create AI models that better reflect the dialects, colloquialisms, and cultural nuances of EU regions.
Meta will notify EU users via in-app notifications and email this week, providing a link to an objection form for those who want to opt out. The link will also be available in Meta’s privacy policy. This move follows Meta pausing its AI-training plans in Europe last year after being asked to do so by Irish regulators.
Meta claims that training AI on EU user data will enable the creation of models that reflect regional characteristics, including dialects, colloquialisms, and hyper-local knowledge. The company emphasizes the importance of this for multi-modal AI producing text, voice, video, and imagery. This development is part of Meta’s broader effort to enhance its AI capabilities.
This announcement is not Meta’s first foray into using user data for AI training. Last year, Meta announced it would start training its AI models on data from British users, who, like EU users, have robust data protection guarantees. Moreover, Meta admitted to having trained AI using all text and photos publicly posted by adult Facebook users since 2007.
