Tech companies have begun to rely on artificial intelligence and machine learning in order to help sort through and identify content on its platforms. But, in order for AI to do its job, an actual person has to teach it first.

A recent report from Reuters has found that Facebook hires contract workers to label posts from both of its platforms for AI systems — including private ones.

In April 2018, Facebook launched its project with Indian outsourcing firm Wipro, who has employed up to 260 workers to annotate posts. According to Reuters, workers sort posts according to five “dimensions,” including the subject of the post. For example, workers have to answer is it a selfie? A picture of food? What is the occasion? What is the author’s intent?

Workers at Wipro look at a wide range of content from both Facebook and Instagram. That includes statuses, videos, photos, shared links, and Stories. To ensure accuracy, everything is checked over by two workers, and workers run through about 700 items each day.

Although Reuters’ report focused on an Indian firm, Facebook confirmed with the outlet that labelers in Timisoara, Romania and Manila, the Philippines are all a part of the same project.

Using people to sort through posts isn’t shocking by itself. However, Facebook could be in violation of the European Union’s General Data Protection Regulations (GDPR).

The GDPR sets strict limits on how companies can collect and use personal data. In addition, the regulations allow people to request that companies delete data about them — and those companies have to follow.

“Facebook said it has technology to routinely sync labeled posts with both deletion requests and changes to content privacy settings,” Reuters reported.

The company says both its legal and privacy teams approve of the data-labeling projects. Facebook also told Reuters that it recently introduced an auditing system to make sure privacy and parameters are working accordingly.

However, Facebook doesn’t have a great history with data. In preparation of a huge FTC fine over the company’s Cambridge Analytica scandal and general lack of privacy practices, the company recently set aside $3 billion.