META has confirmed its plans to use content shared by adult users in the EU to train AI models.
The announcement follows the recent launch of Meta AI features in Europe, and aims to increase the capabilities and cultural relevance of AI systems for the diverse population of the region.
In his statement, Meta wrote: “Today, we are announcing plans to train AI in the meta using public content, including public posts and comments shared by adults on EU products.
“People interactions with meta-AI, such as questions and questions, are also used to train and improve models.”
Starting this week, users of Meta’s platforms in the EU (including Facebook, Instagram, WhatsApp and Messenger) will receive notifications explaining how much data is being used. These notifications are delivered both in-app and via email, detailing the types of public data associated with them and linking to the challenge form.
“We have made this objection easy to spot, read and use. We respect all the form of objection that we have already received and the newly submitted forms,” explained Meta.
Meta explicitly made it clear that certain data types remain off limits for AI training purposes.
The company says it “does not use private messages from people with friends and family” to train generative AI models. Additionally, public data associated with accounts belonging to users under the age of 18 in the EU will not be included in the training dataset.
Meta wants to build AI tools designed for EU users
Meta positions this initiative as a necessary step in creating AI tools designed for EU users. Last month, Meta launched an AI chatbot feature on its European messaging app, framing this data usage as the next step in improving its services.
“We believe that we have a responsibility to build AI that is not only available to Europeans, but is actually built for them,” the company explained.
“It means everything from dialect and colloquialism to hyperlocal knowledge and the clear way in which different countries use humor and irony in our products.”
This becomes more and more appropriate as AI models evolve with multimodal features that span text, audio, video and images.
Meta also positions behavior in the EU within the broader industry landscape, pointing out that training AI for user data is a common practice.
“It is important to note that the types of AI training we are doing are not unique to the meta, nor are they specific to Europe,” reads the statement.
“We’re following examples set up by others, including Google and Openai. Both train AI models using data from European users.”
Meta further argues that the approach is open and outperforming others, saying, “We are proud that our approach is more transparent than our counterparts in many industry.”
Regarding regulatory compliance, META referenced prior engagement with regulators, including delays launched last year while awaiting clarification of legal requirements. The company also cited a favorable opinion from the European Data Protection Commission (EDPB) in December 2024.
“We welcome the opinions provided by EDPB in December, which confirmed that our original approach meets our legal obligations,” Meta wrote.
Broader concerns about AI training data
While meta presents its approach in the EU as transparent and compliant, the practice of using the vast range of public user data from social media platforms to train large-scale language models (LLMs) and generation AIs continues to raise major concerns among privacy advocates.
First, the definition of “public” data can be controversial. Content published on platforms such as Facebook and Instagram may not be posted in the hope that it will become raw materials for training commercial AI systems that can generate whole new content or insights. Users may publicly share personal anecdotes, opinions, or creative works within perceived communities without assuming large, automated analyses by platform owners.
Second, the effectiveness and fairness of the “opt-out” and “opt-in” systems is controversial. Many times, after receiving notifications buried among countless others, when you hold users accountable for actively opposed, questions about informed consent are raised. Many users are unable to review, understand, or act on notifications, and data may be used by default rather than explicit permission.
Third, the issue of inherent bias is looming large. Social media platforms reflect and sometimes amplify social biases such as racism, sexism, and misinformation. This AI model has been trained to risk learning, replicate, and even scale these biases. While companies employ filtering and fine-tuning technologies, eradicating biases absorbed from billions of data points is an immeasurable challenge. Trained with European public data, AI requires careful curation to avoid perpetuating stereotypes and harmful generalizations about the very culture it aims to understand.
Furthermore, questions surrounding copyright and intellectual property persist. Public posts often contain original text, images, and videos created by the user. Use this content to train commercial AI models. This enters dark legal territory with regard to ownership and fair compensation, generating competing content and deriving value from it.
Finally, while Meta emphasizes its transparency compared to its competitors, the actual mechanisms of data selection, filtering, and specific influence on model behavior often remain opaque. Truly meaningful transparency includes deeper insight into how certain data affects AI output and how safeguards to prevent misuse and unintended consequences can affect.
The approach taken by Meta in the EU emphasizes the placement of huge value technology giants in user-generated content as fuels the fast-growing AI economy. As these practices become wider, debates surrounding data privacy, informed consent, algorithm bias, and ethical responsibility of AI developers will undoubtedly be stronger in Europe and beyond.
(Photo of Julio Lopez)
See: Apple AI emphasizes privacy with synthetic and anonymous data
Want to learn more about AI and big data from industry leaders? Check out the AI & Big Data Expo in Amsterdam, California and London. The comprehensive event will be held in collaboration with other major events, including the Intelligent Automation Conference, Blockx, Digital Transformation Week, and Cyber Security & Cloud Expo.
Check out other upcoming Enterprise Technology events and webinars with TechForge here.
