Meta Platforms is preparing to start training its own artificial intelligence models using data from users of its apps in the European Union. This data includes public posts and comments on platforms like Facebook and Instagram, as well as user interaction history with the Meta AI assistant.
According to Meta, personal messages between users and their friends or family members will not be used in the training process. Additionally, only the data of adult users will be involved. Notifications will be sent out to EU residents this week, including a form that allows individuals to opt out if they do not wish for their data to be used to train neural networks.
Focus on Regional Specificity and Transparency
Meta emphasizes that using EU-based user data will help its AI models better account for regional and cultural differences. This, the company states, is especially valuable for the development of multimodal neural networks. These systems are capable of generating not only text responses to user queries but also images and videos, which increases the relevance of localized AI capabilities.
Although Meta initially intended to start this process last year, it was paused due to concerns raised by the regional industry regulator. The company has since adjusted its approach and is now reintroducing the initiative, aiming to align with local requirements. We’ll keep you updated as more integrations become available and as the rollout progresses across EU territories.
Previous Use of Public Facebook Data
Meta’s move follows a similar attempt in the UK, where it also announced plans to train AI models on local user data, reminds NIXSolutions. However, it appears unlikely that the company will have access to the same volume of data as before. In 2023, Meta revealed it had used publicly shared Facebook posts and images — dating back to 2007 — to train its AI. This past usage underlines the scale at which Meta has previously collected and utilized user-generated content to improve its AI technologies.
As this new wave of AI development continues, EU users are encouraged to review the upcoming notifications carefully and consider their preferences regarding the use of their data in AI model training.