Art, AI & Infringement: A Copyright Conundrum

Culture

from the beep-boop-beep dept

I don’t want to waste any space with a long introduction, other than to say it’s always incredibly frustrating when artists come up with inventive new ways to produce artwork, only to have those efforts met with stupid intellectual property issues. Experimentation is key to the artistic world and we’ve begun to see how artists are incorporating technology into what they produce. This should be exciting, but all too often that excitement is plagued by legal issues.

A case in point of this would be Canadian artist Adam Basanta, who has come up with a bonkers and very cool method for both producing machine-generated art and then validating that art for human consumption by comparing it to real-world artwork made by us lowly apes. Let’s start with his setup.

Broadly, Basanta’s machine has two stages: creation and validation.

Creation happens with a hardware setup that Basanta likens to a Rube Goldberg machine: two computer scanners tipped on their sides and pointed face to face, endlessly scanning each other, and the results – influenced by shifts in the room’s lighting, randomized settings and an automatically moving mouse – are interpreted by a computer and turned into colourful abstract pictures.

The second stage is validation. Another computer running a custom-built program automatically checks each image against an online database of real art made by human hands. If the machine-made image is similar to one that has been human-made, the computer dubs it a success and keeps it; if there is no match, the image is deleted forever.

If that doesn’t get your heart beating a little faster, you simply don’t care about art. This setup is, at the very least, incredibly interesting, and Basanta’s method for validating whether the art produced by the machines is good enough for human consumption or not kicks the interest level into overdrive. His setup generates something like a thousand images a day, with a tiny fraction of that being deemed worthy of retention. The whole thing was good enough to warrant an art exhibit in Canada and Basanta has featured many of the images on his website as well.

And that’s where the trouble started. Artist Amel Chamandy has alleged that Basanta violated her copyright on a piece she created called “A World Without Trees”, as well as the trademark rights she has on her own name. Both claims stem from one of the pieces Basanta’s machine setup used to validate its own artwork against and the naming convention it used to denote the new pieces it created.

In June, someone – it’s not clear if it was Chamandy herself or someone who works with her – did a Google search for her name and the name of a 2009 wall installation she made called Your World Without Paper.

The first result in the Google search, according to documents filed in court, was Chamandy’s website. But the second and third results pointed to Basanta’s website, because his machine had named one of its own pictures after one of hers. The offending image, some magenta lines on a field of indigo, is called: 85.81%_match: Amel Chamandy “Your World Without Paper”, 2009.

The trademark claim rests solely on the name of the file including Chamandy’s full name. It’s a silly argument for trademark infringement as the whole point of including the name is to weigh the new art piece against her specific work, which necessarily involves anyone viewing these pieces being informed that they are not the work of the original author. The whole purpose of the validation process is to show what differentiation remains between the new piece and the human-made example. That’s not trademark infringement. It’s not really even close.

As for the copyright portion of this, it’s important that you not be fooled by the percentage the machine setup notes in the validation process. You might think that an 85% match would mean the two images are very similar and would share a ton of features that would link the two in the viewer’s mind. That’s not even close to being the case, as you can see just how different the two images are below.

If that looks like copyright infringement to you, you need your head examined. Indeed, the entire setup here is defined by the fact that this is a totally independent creation — and the “validation” process only serves to highlight that there is no copying. Indeed, the idea that independent creation is a defense against copyright goes back ages, and this is quite obviously an independent creation. The only reason the other artwork is mentioned at all, because it’s the literal coincidence that the computer judged these images similar that leads to the name being mentioned. Judge Learned Hand famously wrote:

… if by some magic a man who had never known it were to compose anew Keats’s Ode on a Grecian Urn, he would be an “author,” and, if he copyrighted it, others might not copy that poem, though they might of course copy Keats’s…

This is a case where “some magic” took place, and one artist “composed anew” something that a computer (but no human eye) judged to have a decent level of similarity to another’s work.

Were her name and the name of her work never mentioned on Basanta’s site, she simply never would have noticed. Nor would anyone else. Ever. And, yet, because Basanta’s entire project centers around pointing out the kind of quality his machine setup can produce in artwork by comparing it to real-world creations made by humans, suddenly Basanta is mired in intellectual property claims.

And that’s what sucks more than anything. One artist suing another, on incredibly specious grounds, is a betrayal of how art is created in the first place. If anything, Basanta was crediting Chamandy and pointing people toward her wider works by doing things the way he did. And this is the thanks he gets, because copyright.

Filed Under: adam basanta, ai, atel chamandy, copyright, matching, scanners, trademark