Meta to use EU user data for AI training

Meta has announced plans to use data from EU users of Facebook and Instagram to train its AI models.

This data will include public posts, comments, and interactions with Meta AI, but will exclude private messages between users.

EU users will start receiving notifications this week through in-app messages and emails.

Each notification will include a link to an objection form for users who wish to opt out, which is also available in Meta’s privacy policy.

Meta paused similar plans last year after pressure from Irish regulators but says this move supports cultural relevance in AI.

🔗 Source: The Verge

🧠 Food for thought

1️⃣ Metadata collection raises privacy concerns beyond visible content

Meta’s announcement focuses on collecting user-visible content like public posts and comments, but doesn’t address the potential collection of invisible metadata that often accompanies digital content.

Metadata, defined as “data about data,” can reveal significantly more information than the content itself, including patterns of behavior, associations, and even location data that users might not realize they’re sharing 1.

This hidden layer of information has historically exposed sensitive details, as demonstrated when German politician Malte Spitz’s six months of phone metadata revealed detailed patterns of his movements and associations 1.

For EU users, this raises questions about what additional information beyond the visible content Meta might collect during AI training, especially since metadata has been documented to pose specific privacy risks even when the primary content appears benign 2.

The distinction between content (what users consciously share) and context (metadata automatically generated) becomes crucial when evaluating the full privacy implications of Meta’s data collection plans 3.

2️⃣ Tech companies increasingly adopt opt-out rather than opt-in for AI training data

Meta’s approach of notifying users about data collection with an opt-out option rather than requiring affirmative consent follows an emerging pattern among major tech companies developing AI.

Companies including Google (for Gemini), OpenAI, Microsoft (for Copilot), and X (formerly Twitter) have all implemented similar approaches, where user data is automatically included in training datasets unless users take specific action to opt out 4.