Flickr Commons and Facial Recognition

Comments on @angeldan’s Flickr post reacting to the NYTimes article on Flickr Commons photos being used to train facial recognition software and the initial reaction of how dare they use our photos for this?

FWIW, it was Yahoo that did this, and they no longer own Flickr.

Also, if they only used photos that were licensed for creative commons or commercial use…well, that’s what those licenses are for, right? Pre-emptively giving others permission to use the work under specific terms without them having to ask?

A. The thing is, you have to make an effort to choose a Creative Commons license for your photos. And the licenses are short, and go out of their way to explain what they do. It’s not the default, and it’s not 40 pages of legalese that you’re required to click through just to use the service. Now, if Yahoo used photos that *weren’t* licensed to allow reuse, then that’s another story.

More concerning, IMO, is whether the *subjects* of the photos would have agreed to this use. Not an issue with identifying trees, birds, cars, etc. – the original Flickr Commons collection wasn’t specific to people – but definitely an issue with the face recognition projects, and that’s where things like the Illinois biometrics law come in.

B. Yeah, the uses that the tech has been put to are in many cases creepy, unethical, and dangerous. That’s always a problem with tech. We often don’t anticipate the ways people will abuse it, and that’s something that we need to do more of. That’s a *huge* discussion in the tech world right now, and one that needs to be examined more closely, from as many angles as possible.

But is this *Flickr’s* breach of trust?
1. It was Flickr’s former parent company that made the original collection, using images that users had specifically assigned creative commons licenses.
2. It was a team at the University of Washington that narrowed it down to the images of faces, created the MegaFace dataset, and encouraged groups building facial recognition technology to use it for training and testing.
3. It’s still other companies that have been actually using the tech.