Meta has announced plans to resume training its AI models on public content from platforms like Facebook and Instagram in the European Union (EU) starting this week. The move comes after a pause in response to regulatory concerns over data privacy, particularly due to the General Data Protection Regulation (GDPR).
Meta had previously delayed its AI model training in the EU, following pressure from data regulators, particularly the Irish Data Protection Commission (DPC), which enforces GDPR rules. The company had suspended its plans in June 2024 after concerns were raised about using user-generated content for training without a clear legal basis.
However, in recent months, Meta has addressed these concerns and has now received clarification from the European Data Protection Board (EDPB). In its blog post, Meta confirmed that the EDPB’s opinion in December 2024 affirmed the company’s approach, allowing them to move forward with training AI models using public content from EU users.
Training AI on Public Content and Interactions
Starting this week, Meta will use posts, comments, and other public interactions from Facebook and Instagram users in the EU to train its AI models. This includes using users’ interactions with Meta AI to further enhance the capabilities of the models.
Meta emphasized that only publicly available data will be used for this purpose. The company clarified that private messages and data from users under the age of 18 will not be involved in training the AI models.
“Since last year, we’ve been waiting for the regulatory landscape to be clarified. We believe we have a responsibility to build AI for Europe by using a variety of local data to ensure our models understand the unique nuances of European communities,” Meta explained.
Transparency and Opt-Out Options
Meta will notify EU users about this change via in-app and email notifications, providing them with a link to an opt-out form for those who do not wish to have their public data used in training. The company has assured users that all opt-out requests, both past and future, will be respected.
The move to start training AI models on public data in the EU follows the steps of other tech giants like Google and OpenAI, which have already used European user data for their own AI model development.
Impact on European Communities
Meta emphasized that its generative AI models will benefit from exposure to diverse European data, enabling the models to better understand regional dialects, cultural nuances, humor, and colloquialisms. This, according to Meta, will ensure a more localized and relevant AI experience for users in Europe.
Ongoing Regulatory Oversight
While Meta has moved forward with its plans, the DPC continues to investigate how AI models, including those from other companies, are trained. Recently, the DPC also announced an investigation into xAI’s training of the Grok AI model.
With the EU regulatory hurdles addressed, Meta is now poised to harness public content to improve its AI models, offering more localized and personalized experiences for European users. This step signals a broader trend in the tech industry, as companies like Meta, Google, and OpenAI navigate complex data privacy regulations while striving to advance AI technology.
