Meta will also train AI on data from Europeans from the end of June

Meta also appears to be training its AI systems on data from European users. This will happen from the end of June, according to emails that various users have received. Meta says it has a 'legitimate interest' in this regard.

Meta writes this in an email to users, including a number of tweakers. “We are preparing to expand our AI at Meta experiences to your region,” the company wrote. In practice, these 'Meta-experiences' are chatbots and other generative AI that are incorporated into services such as Instagram, WhatsApp and Facebook. These services are trained on visible user data, such as posts and comments. This has already happened in other countries, but not yet in Europe due to stricter privacy rules.

That is now changing, Meta warns. The company points to a new privacy policy, which will become active on June 26 this year. It states that Meta will also use data from Europeans to train AI models.

Meta relies on legitimate interest, one of the bases that companies can use to process data under the GDPR. That is striking, because Meta has been reprimanded more than once by supervisors because of that basis. The company used this for years to collect data for advertising purposes, for example, but after various fines and other sanctions it decided last year to rely on consent instead of legitimate interest. When using data for generative AI, the company has opted for that controversial basis.

Meta does write that users can object to the use of their data. This is an obligation under the GDPR. The company writes this both in the privacy policy and in an email to users. To do this, users must complete a form where they must provide a reason why they do not want to participate.