IBM Used Creative Commons Photos from Flickr to Train AI Facial Recognition

By this point we really shouldn’t be surprised when privacy is disregarded. It’s become more than apparent that businesses don’t feel the need to protect our privacy. Nevertheless, photographers are surprised that IBM stretched the Creative Commons license of photos they posted to Flickr to train their artificial intelligence facial-recognition systems.

IBM Used Flickr’s CC Photos

There’s no other way to look at this other than a company taking liberties that no one expected them to. When you share a photo under a Creative Commons license you have certain expectations, and those expectations don’t include IBM using them to develop technology.

Flickr is one of the larger sites on the Web that allows users to add photos and apply a Creative Commons license. Photos and other media of people, objects, locations, etc., are shared for free without a copyright. The licenses can be commercial or non-commercial and can require anyone using the photo to provide attribution to the source.

Academic researchers with IBM took these free photos shared under the Creative Commons license and used them in their Diversity in Faces program. The faces in the photos were annotated with factors such as sex, age, measurements, etc. It was done with the approval of the company’s legal team and was done to counter bias that can affect recognition and fairness.

The question of commercial or non-commercial can be debated here. These photos are being used to develop technology that will be part of marketable software or devices, but the faces themselves aren’t actually being marketed. Additionally, there is no way to attribute these photos. And the photographers behind the photos are receiving no compensation for the photos’ usage.

“None of the people I photographed had any idea their images were being used in this way. … It seems a little sketchy that IBM can use these pictures without saying anything to anybody,” offered PR firm SharpOrange’s Greg Peverill-Coni, an executive, to NBC News.

IBM does not feel it did anything wrong. They feel they have maintained the privacy of the people whose faces were used to develop their AI 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,” said IBM spokesman Saswato Das in a statement.

news-ibm-flickr-photos-blue-eyes

“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.”

The lesson here is for photographers to know that photos being shared could have that license stretched. Sometimes on this site you’ll see our annotation at the bottom of articles noting “Image Credit” where we provide proper annotation.

We will use CC-licensed photos from sites like Flickr or Wickimedia Commons, but we provide annotation and only use photos cleared for commercial use. Or we use our own photos, such as with screncaps, or photos that are provided on other sites under public domain, like the ones I have used in this article.

SmugMug Chief Executive Don McAskill, whose company acquired Flickr after IBM used the photos defended the usage of the photos. “People didn’t have to opt in to the dataset because they had already opted into the Creative Commons license. They took action. This is the way licensing works. It’s also the magic that enables artists and scientists all over the world to create and invent using CC-licensed works,” he stated.

However, Chief Executive Ryan Merkley of the Creative Commons organization said, “Our tools were built to solve for copyright, and they do that well. But copyright isn’t a good tool to address privacy, or research ethics, or surveillance AI.”

User Beware

This makes me remember when I worked for a printing company as a computer typesetter. The others in the art department were refining their process of shooting photos to be printed. They used the photo of a local real estate agent because he was balding, and they wanted to use the fine hairs on the top of his head to be sure they weren’t lost and remained sharp.

His picture was used for years every time they wanted to refine the system. It wasn’t illegal. No one was marketing his photo. But people were staring at his image and those hairs on his head, something he never realized would be done when he posed for the photo that would be used in advertisements for his company.

It’s not illegal. Perhaps it’s more of a moral issue. Some will see it as okay, and some won’t. What do you think? Is this a moral issue? Is this something a company like IBM should be allowed to do? Add your thoughts on Creative Commons licensing of online photos to the comments below.

3 comments

  1. Inasmuch as I am a security and privacy fanatic, this non issue is a tempest in a teapot. I would like to be proven wrong by somebody proving that IBM caused them any harm by using the pictures.

    “those expectations don’t include IBM using them to develop technology”
    Why not? Remember, anything that is not explicitly forbidden is allowed. No names or any other data was revealed or is even known. People in the pictures are recognized only by their family and friends. Pictures are in the public domain therefore there is no expectation of privacy either express or implied. Expectations are only oour hopes. They are not legally binding.

    I would not be at all surprised if my picture was one of those that IBM used. However, I do not feel in any way diminished if my picture was used nor do I feel as if my privacy was invaded, rudely or surreptitiously.

    BTW – some time ago there was a big to-do about the AI Facial Recognition algorithm being biased because it recognized white faces more accurately than those of people of color. If I remember correctly, the database of faces consisted of only the employees of the company. Unfortunately for the results at the time, the employees were mostly white. By using the Flickr pictures, IBM has rectified that shortcoming, the database reflects the population better, the algorithm has orders of magnitude more records to use so that it can give much more accurate results. Unfortunately, bias concerns have been replaced by privacy concerns in people’s minds. Why don’t we just admit that we DO NOT WANT AI Facial Recognition to exist.

  2. Article: “Photos and other media of people, objects, locations, etc., are shared for free without a copyright.”

    No. All media we create has copyright protection automatically unless we specifically put it into the public domain. Granting a license, in this case a Creative Commons license, does _not_ eliminate the copyright.

    dragonmouth: “Why not? Remember, anything that is not explicitly forbidden is allowed.”

    No, it is the opposite. Our copyright forbids any use, except some unhelpfully vague “fair use” provisions. The license we apply is what grants permission for certain uses. I’ve not carefully read the CC licenses that were applied to these photos (nor am I an attorney), so I don’t know if this was covered by them or not.

  3. I agree with dragonmouth on this – if someone is willing to have their picture used in an advertising campaign, or where it’s already available for anyone to see on Flickr, I don’t see why there’s any issue with IBM using them to make their solution better. They could have printed out all the photos and plastered them on the wall in their buildings, and nobody would have any legal standing to complain, based on the CC license. Why is using them for research and AI improvement any “worse” than that?

Leave a Comment

Yeah! You've decided to leave a comment. That's fantastic! Check out our comment policy here. Let's have a personal and meaningful conversation.

Sponsored Stories