2 min

Tags in this article

, , ,

IBM is said to have offered a database of nearly a million photos, without the consent of the people who took these photos. The photos were offered to train facial recognition technology. These were photos with clear faces on them.

The photos were taken from Flickr and provided with comments to describe the face on the photo. The collection was portrayed by IBM as a way for researchers to reduce prejudices in facial recognition.

However, some of the photographers of those images did not know that their photos were in that database, NBC News discovered. The American news medium informed the photographers about this. “None of the people I photographed knew that their images were used in this way,” says Greg Peverill-Conti, a PR leader with over 700 photos in the collection.

Other investigations

The photos were placed on Flickr under a Creative Commons license, which means that in some cases they can be used for many purposes. Flickr’s owner, Yahoo, made a collection of 100 million photographs available to researchers under that license in 2014. So IBM took pictures with clear faces and made them available in the Diversity in Faces-dataset.

Moreover, IBM is not the only company that uses publicly available photos on the Internet in such a way. Dozens of other research organizations collected photos to train facial recognition systems. According to Google Scholar, hundreds of academic papers have been written on the basis of such collections of photographs.

IBM itself states that people who are in the dataset can ask for their photos to be deleted. NBC News, on the other hand, says that this is difficult, because links to the photos to be removed have to be emailed. However, there is no public tool or list to see if someone is in the dataset. NBC has now made its own tool for this purpose.

This news article was automatically translated from Dutch to give Techzine.eu a head start. All news articles after September 1, 2019 are written in native English and NOT translated. All our background stories are written in native English as well. For more information read our launch article.