Meta has resumed training its AI models in the European Union, using public data from adult users. The company paused this effort last year due to concerns raised by Irish regulators over data privacy. Following approval from the European Data Protection Board, Meta will now use user posts, comments, and interactions to improve its AI. The aim is to help the models better understand and reflect European cultures, languages, and history.
The data collection will exclude private messages between individuals and the data of users under the age of 18. Meta has begun notifying users in the EU, via email and in-app messages, about the types of data being used. These notifications will also explain how this data collection will improve the AI and user experience.
Users are given the option to opt out of having their data used for AI training purposes.
Meta emphasized that it would honor all objections to data usage that have been previously submitted, as well as any new ones. This move comes shortly after Meta launched its Meta AI service across Facebook, Instagram, WhatsApp, and Messenger in the EU. The company pointed out that other tech giants, like Google and OpenAI, have followed similar practices of using European user data to train their models.
Meta’s decision aligns with this trend while ensuring compliance with EU data protection laws.
In parallel with Meta’s announcement, Apple shared how it uses techniques like differential privacy and synthetic data. These methods help Apple improve its features like Genmoji, Image Playground, and Writing Tools in Apple Intelligence. The company ensures that user privacy is not compromised by sending device analytics to improve services. Apple’s approach demonstrates how AI development can be advanced without sacrificing privacy, a practice Meta is now aligning with.
Reference:
