Meta gets EU green light to train AI on public user data
- Marijan Hassan - Tech Journalist
- 23 hours ago
- 2 min read
Meta has secured regulatory approval in the European Union to use public user content for training its AI systems — a significant step toward expanding its generative AI capabilities across Europe.
Starting this week, users in the EU will receive in-app and email notifications informing them that their public posts and comments may be used to train Meta's AI models. The notifications will explain what data is being used, why it's being used, and how users can opt out if they choose. Meta says it will honor all opt-out requests, including ones already submitted.
This marks a major regulatory milestone for Meta, which previously held off on training its large language models in the EU while legal uncertainties were clarified. In December, the European Data Protection Board affirmed that Meta's approach was legally sound. Since then, Meta has worked with Ireland’s Data Protection Commission to move forward with a strategy it describes as “more transparent than many of our industry counterparts.”
Scope of training data
Meta is limiting the scope of the data it uses. Private messages, content from users under 18, and non-public data will not be included in training datasets. Instead, the company will rely on publicly available posts, comments, and interactions with its AI tools to fine-tune its models.
Meta says the move is essential to ensuring its AI systems reflect the rich cultural, linguistic, and social diversity of Europe.
“We believe we have a responsibility to build AI that’s not just available to Europeans, but is actually built for them,” the company said in a statement. “That means understanding dialects, colloquialisms, humor, sarcasm — and how those elements vary across countries and communities.”
A more personal AI assistant
The announcement follows Meta AI’s launch in Europe last month, which introduced the company’s chat assistant into messaging apps like Facebook, Instagram, Messenger, and WhatsApp. With regulatory clearance now in hand, Meta is ramping up its efforts to personalize and localize its AI services for the European market.
This strategy puts Meta in line with competitors like Google and OpenAI, both of which have already used public European data to train their AI models. However, Meta claims its opt-out system and public communication make its rollout more transparent.
As generative AI becomes more multimodal, expanding into voice, video, and image generation, the quality and diversity of training data becomes even more critical. By tapping into public European content, Meta is betting that a more regionally grounded AI will better serve its 450 million users across the EU.
