Mark Zuckerberg Chairman and CEO of Meta Platforms (formerly Facebook, Inc.) | Meta Platforms (formerly Facebook, Inc.)
Meta Plans to Expand AI Features in Europe
For years, Meta has been developing the next generation of AI features across its apps and devices. This year, the company plans to expand its collection of generative AI features and experiences, along with the models that power them, to users in Europe.
AI at Meta is already available in other regions, including Llama, an open-source large language model, and the Meta AI assistant. To serve European communities effectively, these models need to be trained on information relevant to Europe's diverse languages, geography, and cultural references. This requires using public content shared by Europeans on Meta’s platforms.
"If we don’t train our models on the public content that Europeans share on our services and others... then models and the AI features they power won’t accurately understand important regional languages, cultures or trending topics on social media," a spokesperson for Meta stated. "We believe that Europeans will be ill-served by AI models that are not informed by Europe’s rich cultural, social and historical contributions."
Meta is following precedents set by companies like Google and OpenAI, which have also used data from European users for training their AI. However, Meta emphasizes that its approach offers more transparency and easier controls than many industry counterparts.
In line with its commitment to responsible development practices that adhere to local laws and regulations, Meta is consulting with the Irish Data Protection Commission. The company has incorporated feedback from this regulator to ensure compliance with EU privacy laws. Additionally, Meta continues working with experts such as academics and consumer advocates.
To inform users about their rights and available controls since May 22nd, Meta has sent over two billion notifications via apps and emails to people in Europe. These notifications include a link to an objection form allowing individuals to opt-out of having their data used for AI modeling.
The company reviewed approaches from industry counterparts while developing these notifications. As a result, it designed its form to be easily accessible within three clicks and readable even at lower literacy levels.
"We are honouring all European objections," said a spokesperson for Meta. "If an objection form is submitted before Llama training begins...that person’s data won’t be used."
Meta clarifies that its goal is not to create a database of individual information but rather identify patterns such as colloquial phrases or local references from publicly shared posts by users over 18 years old. Private messages between friends or family are not used for training AI systems.
Like other companies using Europeans' data for training large language models under GDPR guidelines, Meta relies on 'Legitimate Interests' as its legal basis. This balance aims to process public data necessary for training AI models while respecting user rights.
"To do this work in the EU...we feel a responsibility to build AI that is not forced on Europeans but actually built for them," said a spokesperson.
As Europe stands at the forefront of technological evolution amid debates over data usage policies some activists advocate extreme measures against data use for AI development—positions that do not reflect European law according to Meta.
"As one of the most influential regions in the world...Europe has potential competitive leadership in AI innovation," stated a representative from Meta emphasizing Europe's opportunity amidst global advancements.
Meta aims for Europe’s active participation during this pivotal technological moment ensuring equal access reflecting regional culture humor history contributing towards community growth driven by innovative technology.
