The AI Copyright Nightmare

Jul 16

Recently, Figma unleashed its new AI features to the world – only to get rid of them shortly after. According to an article by Sarah Perez for TechCrunch, it all started when a person by the name of Andy Allen used the new “Make Designs” AI feature to make a weather app. But the tool ripped off Apple’s weather app three times.

Figma quickly realized their mistake. Dylan Field, the CEO, took responsibility, noting that he rushed his team to develop the feature prior to their Config conference.

Just like that, we now have another example of an AI horror story.

As if AI hasn’t been horrific enough.

I’m never going to argue that designers shouldn’t use AI, because I am very much on the side that they should. But all designers should know about AI’s limitations and potential dangers.

A Copyright Nightmare

Let’s note that Gen AI can be – scratch that, it is – a copyright nightmare. In order for AI to generate anything, it needs to be trained on existing stuff.

And where are companies getting this stuff to train their AI’s on? Apparently, that was Figma’s issue. The company itself didn’t train their AI; they got someone else to do that.

Which leads to the second point, which is that AIs are not made equal. According to an article by Beth Nichols for CreativeBloq, Adobe Firefly is “is said to be one of the most ethical AI image generators out there.” Compare that to Midjourney, which according to an article by Calvin Wankhede for Android Authority, “was trained on existing image samples, including art from various sources, to generate brand-new pictures. Some artists believe that AI image generators have infringed on their copyright by using their work for training.”

To summarize: Firefly uses inspiration, while Midjourney samples without credit. This is where things get touchy – what Gen AI images, if any, can designers actually use as a part of their work? And is it safe for designers to say that images generated using AI are theirs, or is that actually copyright infringement?

An Issue of Attribution

I watched a really interesting video by independent journalist Cleo Abram in which she explains the use of AI in music and how it relates to copyright issues. In the video, she presents a chart comparing taking inspiration for artists vs copying, and how artists either pay or not pay in each case. The video explains it better than me, however, so feel free to take a couple minutes to watch it.

There are instances in which artists sample music from another artist, in which they copy and pay. If they sampled, but didn’t pay, that’s stealing.

Now when it comes to inspiration, typically, that’s not paid back to the original artist. I’m inspired by a lot of graphic design professionals, but none of them get compensation from my work. But can one be inspired by existing artists, and allow those artists to be paid? It’s quite possible.

So where does this Gen AI artwork fall into these categories? One can argue that AI uses existing images as inspiration. Another can argue that it falls into the copying category. But we fail to recognize the other category, the compensation piece. That piece only identifies the bottom line when it comes to AI attribution: who gets the original credit?

What’s the Answer?

In an article for the Center for Art Law, Atreya Mathur writes that “Copyright cannot belong to the AI itself. Section 306 of The Copyright Act protects ‘original works of authorship,’ which implies a human hand in the process.” This means that in order for an artistic piece to have copyright protection, it has to have been made by an actual human being. A machine-generated image can’t have copyright protection behind it – instead, the companies behind them could be in a case of copyright infringement. According to Mathur, “since copyright law does not protect AI-generated artwork, it is likely neither the AI nor the AI company has any rights in the image.”

Designers should proceed with caution when using Gen AI. Sure, it is great for quickly visualizing an idea or creating an image that you need, but it cannot be protected by copyright. And based on this research, it’s not clear who has the copyright.

These are shallow, uncontested waters – and we need to keep an eye out for the sharks.