UK Edition
IBM criticised for collecting social media photos for facial recognition research
IBM has been criticised for harvesting photos from internet users’ social media profiles without their knowledge to train its facial recognition algorithms.
The American technology company “scraped” around one million photos of members of the public from the photo sharing website Flickr and used them to conduct research into bias in facial recognition technology.
Users were unaware that their photos had been collected for the project, and had no way of knowing that photos taken from their personal accounts were included in the data set, an investigation from NBC found.
IBM said the photos were only used for research purposes, not to improve its commercial facial recognition technology, and that all the photos used had a “creative commons” licence that allows photos to be taken and re-used. The images were only shared with verified researchers, it added.
Tiffany Li, of Yale Law School's Information Society Project, said the project was dubious. “Even if IBM was legally in the right, this project is ethically questionable. They should have addressed these concerns head-on,” she said.
Researchers developing facial recognition technology used for policing and security require huge sets of real-life photos in order to “train” the artificial intelligence algorithms behind the technology.
Social media profiles and online photo storage sites have become a ripe source of images, but users are typically unaware that their photos may be used for this purpose, even if they had made their images public and agreed that others could re-use their work.
IBM said it would remove users’ photos from the database if they had requested it. However, it is not sharing the collection publicly, meaning members of the public could not know if their photos are being used.
“Individuals had no way of knowing IBM would use photos of them, so they could not have consented to that,” Ms Li said. “But they may have consented to being subjects of the photos in the first place.”
Am IBM spokesman said: "IBM has been committed to building responsible, fair and trusted technologies for more than a century and believes it is critical to strive for fairness and accuracy in facial recognition.
"We take the privacy of individuals very seriously and have taken great care to comply with privacy principles, including limiting the Diversity in Faces dataset to publicly available image annotations and limiting the access of the dataset to verified researchers. Individuals can opt-out of this dataset."