Meta plans to use European user data to train AI models

Meta, parent company of Facebook and Instagram, has announced its renewed intention to use public posts, photos and comments from European users to train its AI models. The announcement came via a company blog post. In the coming week, users in the EU will receive notifications explaining the plan and how to opt out.

This is Meta’s second attempt. The company previously withdrew a similar plan in mid-2024 following pressure from European privacy regulators. This time, Meta says it will only use public content from adult users, and will provide a form to file objections. Private WhatsApp messages are excluded from the data collection.

Meta argues that incorporating European user data is necessary to create culturally nuanced AI products—for example, to capture local humour and linguistic context. Without such data, the company says, its AI tools would be “second-rate.”

The broader practice of using public online data to train AI has drawn increasing criticism. During the early AI boom, companies scraped vast amounts of content—ranging from news archives to social media and art—without consent. Meta previously faced backlash for using texts from public e-book databases without author permission.

Meta’s relationship with the European Union has been strained for years over privacy issues. In 2023, the company was fined a record €1.2 billion for transferring European user data to the United States.

Users who do not want their public posts used for AI training can opt out via the notification or email. Meta has said that it will honour both new and previously submitted objections.

Read the full article on the website of de Volkskrant.