Meta is planning to train its AI with EU users’ data

Just a heads up, if you buy something through our links, we may get a small share of the sale. It’s one of the ways we keep the lights on here. Click here for more.

Meta has announced plans to start training its artificial intelligence systems using data from adult users in the European Union. 

This includes public posts, comments, and any chats users have had with Meta’s AI assistant, but the company says it will not use private messages exchanged with friends or family. 

Only users aged 18 and over are included in this data collection. Meta says this step is meant to improve its AI so it better understands the languages, humor, slang, and cultural nuances of different European countries. 

This is especially important for AI systems that work with text, speech, images, and video. The company claims the goal is to make its AI more relevant and useful to people in Europe. 

Users in the EU will start receiving notifications in their apps and via email explaining what’s happening, and those who don’t want their data used will have the option to opt out by filling out a form linked in Meta’s privacy policy. 

As of now, Meta is still waiting for the green light from regulators, which is why the data collection hasn’t started yet. 

In fact, Meta had paused similar AI training efforts in Europe last year after Irish privacy authorities raised concerns. 

This latest move is similar to what Meta did in the UK, where users also have stronger legal protections around how their personal data is used. 

Even though the company is just now seeking consent from EU users, it already has a vast amount of data. 

In 2023, Meta admitted it had trained its AI on public posts and photos shared by adult Facebook users going back as far as 2007. 

So, while the new data collection may seem small, it’s being added to a massive pool of existing information. However, it’s also raising fresh questions about privacy and consent in the process.

What do you think about user data used for training AI? Do you think you would consent to it? We want to hear your thoughts below in the comments, or via our Twitter or Facebook.