Meta got the green light to train AI on EU user data

Meta is resuming its artificial intelligence (AI) training efforts in the European Union (EU) after receiving regulatory approval, allowing the company to use public data shared by adult users across Facebook, Instagram, WhatsApp, and Messenger. The move comes nearly a year after Meta paused similar efforts due to concerns raised by Ireland’s data protection authority.

What kind of data will Meta collect?

According to Meta, the AI training will rely on content made public by adult users, including posts, comments, and interactions with Meta AI tools. Private messages and data from users under 18 will not be part of the training dataset. The company said the collected data will help its AI models better reflect European cultures, languages, and history.

Starting this week, users in the EU will see in-app notifications and receive emails explaining what data Meta plans to use and why. The notifications will also provide an opt-out form, allowing users to object to having their public content used for AI model training. Meta has committed to respecting both new and existing objection requests.

How did regulators respond?

The European Data Protection Board (EDPB) approved Meta’s plan after confirming that it meets the EU’s strict data protection rules. Meta’s decision follows a wider trend in the AI industry, with companies like Google and OpenAI already using European user data to develop their AI models.

The approval clears the way for Meta to continue expanding its AI capabilities across its platforms in the EU. Last month, Meta AI became available to users in the region, integrated into Facebook, Instagram, WhatsApp, and Messenger for tasks like answering questions, generating content, and improving user recommendations.

Meta AI just landed on WhatsApp, Instagram and Messenger in Europe

Apple takes a different approach

While Meta’s plan focuses on using real-world public data, Apple recently outlined a contrasting strategy. In a new report, Apple explained how it uses privacy-preserving methods like differential privacy and synthetic data generation to improve AI-powered features. This includes tools like Genmoji, Image Playground, Memories Creation, and Writing Tools in its upcoming Apple Intelligence suite — all while limiting personal data exposure.

Apple said users who opt to share Device Analytics data contribute to feature improvement, but with safeguards that prevent the company from directly accessing identifiable user information.

Meta’s AI training rollout in Europe will likely reignite debates over data privacy and user consent, especially as tech companies continue expanding generative AI features. For now, adult users in the EU have the option to opt out if they do not want their public content used for AI training.

Featured image credit