Up until a few years ago when a fashion house was blamed for blatantly copying somebody else's design, they would issue a vague statement explaining that something had gone wrong internally, possibly blaming a random member of the staff (as you may remember from a previous post, Chanel claimed "dysfunctionality within its teams"...), an inexperienced graphic artist or intern.
But, rejoyce, if you have stolen a specific design you may have stumbled upon on the Internet and you are now caught in a copyright infringement case, you may be able to blame...the algorithm.
But let's look at this story from the beginning: a couple of weeks ago, artist @Hannahdouken ran an experiment to test the power, skills and copyright infringement possibilities offered by algorithms. The artist posted in a tweet a request, saying "Can y'all do me a favor and quote tweet/reply to this with something along the lines of 'I want this on a shirt', thank you". The image accompanying the tweet simply stated "This site sells STOLEN artwork, do NOT buy from them!" accompanied by a little heart and a smiling face.
The tweet wanted to prove a theory: automated bots are currently actively looking on Twitter for phrases like "I want this on a shirt" or "This needs to be a T-shirt" accompanied by an image. After detecting such sentences and the images posted with them, they steal the visual artworks and create with them print-on-demand T-shirts (obviously without permission) for a variety of sites.
Followers complied with @Hannahdouken's request and, shortly afterwards, a Twitter bot replied with a link to the newly-created T-shirt listing available on print-on-demand t-shirt service Moteefe (more followed on other similar sites, such as Toucan Style and CopThis).
The debate expanded in the days that followed: Twitter user @Nirbion realised that @Hannahdouken had just proved the theory was right, but there was no punishment for the thieves. So he upped the ante and created an image that infringed Disney's copyright, adding under the image "This is NOT a parody! We committed copyright infringement and want to be sued by Disney. We pay ALL court and tribunal fees."
Again bots detected the image and immediately created a T-shirt inspired by the design. Since then some sites removed the "Not Licensed by The Walt Disney Company" Mickey Mouse image, but more Twitter users created copyright infringing images featuring popular characters, including Pikachu and Super Mario.
Creating quick selling garments using algorithms is not a new practice and perfectly explains the wondrous and instant proliferation of a wide range of items with modern and popular designs, logos and graphics on sites such as Aliexpress (motifs that, in some cases, end up providing inspirations for more expensive and exclusive fashion collections).
In 2011 T-shirt producer Michael Fowler wrote a simple computer code that stated "Kiss me, I'm a ____". The slogan could have been completed using a database of vocabularies and word variations selected by an algorithm.
Fowler's idea dramatically increased sales for his company, but, a year afterwards, the company was betrayed by its own algorithm when the latter started generating disturbing slogans moving from the WWII propaganda phrase "Keep Calm and Carry On" (including "Keep Calm and Rape Them").
In the last few years algorithms have often been employed to create a wide range of customizable items: you can indeed find on Amazon mobile phone covers with rather bizarre images stolen from a database of stock images and showing a woman getting a botox injection, toenail fungus, colostomy, and even the picture of a coffin that seems lifted from a funeral parlour brochure.
You may wonder who may want to buy such objects, but even if the T-shirt or mobile phone case in question would sell only one sample, they would still make money.
In most cases these objects are not produced until they are ordered, so there is no physical inventory, but some of the sites offering the T-shirts are integrated with other platforms which means the products immediately appear on other sites as well.
This means that, usually, the items end up selling more than just one piece, as they appear on Facebook users' feeds and they often seem to speak to the users in a very personal way thanks to hyper-targeted digital advertising that allow algorithms to get to know a user's obsessions, passions and personal details.
This is actually the most intriguing (and scary at the same time...), aspect about such products or about the experiments recently carried out by Twitter users: bots are scanning people's tastes to find a design that they may really want to buy, without paying the artist who created it or without caring about what's written in a slogan or an image, and immediately creating a product with it.
How to protect the artists? At the moment there is no legal solution and the identities of print-on-demand providers are often protected by the platforms they work with.
While collective effort is proving effective at carrying out experiments and at spreading the results (and the hilarity as well...), a way to protect artists who may have posted their artworks on social media and avoid them wasting time and money asking a site to take the bootleg merchandise down, is not to enthusiastically answer their posts saying you want a specific product (T-shirt, mobile phone cover, etc) with that particular artwork on. Bots track text, so you can maybe add an image expressing your enthusiasm about a particular artwork.
As for automated design processes, well, let's hope that fashion houses and designers will not start using the excuse "It wasn't me, it was the algorithm" (even though this sounds like the sort of excuse Jeremy Scott may use...) to explain why a specific print, design or work featured in their collections may look too similar to that created by an independent designer and posted on social media. We often talk about how to get Artificial Intelligence to perform human tasks, but at the moment it looks like we have managed to successful teach to it how to steal from human beings.
Comments