Meta will use your social media posts to train its AI. Europe gets an opt out

How fortunate it is to have privacy regulations that are actually useful in Germany! Meta will now begin training its AI models using everyone’s social media posts, although users in the European Union have the option to opt out, a luxury that the rest of the world won’t have.

The Facebook parent company announced this move today, stating that it aims to bring its machine-learning systems to Europe.

Up until now, Meta has excluded its European user base from its AI training data, likely to avoid legal conflicts with the continent’s privacy regulations. However, despite complaints, Meta is now moving forward with this decision.

“In order to effectively serve our European communities, the models that power AI at Meta must be trained on relevant information that reflects the diverse languages, geography, and cultural references of the people in Europe who will use them,” explained the social media giant.
In order to train our powerful AI models that drive our AI features, we aim to utilize the publicly shared content on Meta’s products and services, as chosen by individuals in the EU.

Considering the potential controversy surrounding AI training with user data in Europe, Meta has taken two measures to address this. Firstly, the term “public content” refers specifically to posts, comments, photos, and other content shared by users on its social media platforms, provided that the users are above the age of 18. Private messages are strictly excluded from the training data.

Meta has also made efforts to inform and give options to European users. Since May 22, billions of notifications have been sent to users, allowing them to decline their content being used for AI training. Users have the freedom to decline without any inquiries, ensuring that their posts will not be utilized for AI model training, both presently and in the future.
The situation is quite different in the rest of the world, as opting out is not a viable option. While it may be too late to opt out for Meta’s LLaMa 3 training data, Facebook and Instagram users outside of the EU are still required to participate in training for future models. It is possible that users outside of Europe may have the option to opt out in the future, but currently, this feature is exclusive to the EU.

It is expected that Meta will face some resistance in using European user data, despite feeling confident in their position. Prior to making a public announcement, the company updated its privacy policy, which led to consumer privacy advocacy group noyb filing complaints across Europe.
Noyb argues that user data collection should require an opt-in approach instead of opt-out, as default settings. Additionally, the inability to completely remove data from an LLM or other AI model may pose challenges due to the European Union’s Right to be Forgotten.

Furthermore, Meta and the EU have a strained relationship, as evidenced by recent investigations launched by the EU regarding child safety and misinformation during the EU parliamentary elections. Although the outcome is uncertain, it is likely that Meta will face challenges from the social network in the future.