The complaint stated that photographs of Brazilian children were used without permission to train artificial intelligence.

Personal photos of Brazilian children were used for artificial intelligence training without permission, reported Human Rights Watch (HRW) yesterday (10). The photos were included in a large database that companies use to feed advanced models.

According to the Human Rights organization, approximately 170 photographs of children living in ten Brazilian states were found: Alagoas, Bahia, Ceará, Mato Grosso do Sul, Minas Gerais, Paraná, Rio de Janeiro, Rio Grande do Sul, Santa Catarina and São Paulo.

See also: How to keep kids safe online

However, the organization states that there may be more images in the database since there are more than 5.8 billion files in the database.

HE The name of the image bank in question is LAION-5BAccording to analysis by Human Rights Watch. According to HRW, the database is used to train popular (unspecified) AI tools.

Personal data of Brazilian children

In the images found, children can be identified and in some cases the names of people in the frame are included in the caption or integrated into the URL of the files.

“In many cases, their identity can be easily traced, including information about when and where the child was when the photo was taken,” Human Rights Watch explains.

You The records in the bank belong to various moments of childhood: Immediately after birth, playing with friends, attending seminars in the classroom, and more. HRW notes that some photos were published on personal blogs or media sharing sites.

The presence of images in the artificial intelligence training database not only violates privacy, but also paves the way for deepfakes. HRW states that at least 85 girls have alleged abuse due to the misuse of photos in fake images.

Other side

In response to HRW’s allegations, LAION, the company responsible for LAION-5B, confirmed that there were personal photographs of children in its database and promised to remove them.