Meta officially announced this Monday that it will resume training its artificial intelligence models using public content from Facebook and Instagram users across the European Union. This move follows a strategic pause prompted by intense regulatory scrutiny regarding data privacy compliance.
Regulatory Hurdles and Compliance
The decision to restart training comes after a period of friction with the Irish Data Protection Commission (DPC). Meta previously halted its plans in June 2024 to address concerns surrounding the General Data Protection Regulation (GDPR), which demands a strict legal basis for processing personal data for machine learning purposes.
According to Meta’s official statement, the company feels confident in its current approach. The move follows an opinion provided by the European Data Protection Board (EDPB) in December, which Meta claims validates its compliance with EU legal obligations.
What Data is Being Used?
Starting this week, Meta will begin incorporating public posts, comments, and interactions with Meta AI into its training datasets. The company emphasized specific guardrails to mitigate privacy concerns:
- Private messages remain excluded from the training process.
- Public data from users under the age of 18 in the EU will not be utilized.
- Users will receive in-app and email notifications detailing the process.
The Opt-Out Mechanism
Meta is providing a clear pathway for users who wish to restrict their data from being used in AI development. Notifications sent to European users will include a direct link to an objection form. The company confirmed it will honor all previously submitted objection forms alongside any new requests processed during this rollout.
Why Localization Matters
Meta argues that training on European data is essential to deliver a high-quality user experience. The company stated that to build AI “built for Europeans,” models must grasp the linguistic nuances, dialects, colloquialisms, and regional humor unique to the continent’s diverse communities.
By moving forward, Meta aligns itself with industry peers like Google and OpenAI, which have already integrated European user data into their respective AI frameworks. Despite this progress, the DPC continues to monitor the sector closely, recently launching an investigation into how xAI trains its Grok model.