Copyright is a complex subject, often abused and rarely understood. It all revolves around the question of who has the right to copy an original work, and how that original work is defined.
Bringing AI into the mix has made it even more complex, because you can’t copyright an idea. To claim copyright, that idea needs to be made into something that can be recognised as something more tangible – for example: a book, a painting, or a photograph. The right to copy the combined idea and output also needs to be legally declared.
This is what makes the AI copyright debate interesting. The person inserting the prompt might have the idea, but they lack the skill to make that into something tangible. For that they’re using AI. So, they can’t claim copyright, can the owners of the AI program?
Some might argue that AI is creating. Now AI may be able to generate images and written content based on a prompt, which is essentially an idea, but nothing it creates is ever original. AI can only reproduce what it’s learnt, and even combining multiple sources to create something different, still doesn’t make it original. And this sparks another controversial debate:
Did the LLMs training AI legally obtain the right to copy the original texts and or images they’re using to create these outputs? Turns out, Open AI claims that would have been prohibitively expensive – says the billion-Dollar corporation. The world should be grateful for the capabilities AI is providing, not get caught up in who owns what…convenient.
Sadly, there are many people that agree with the sentiment. Probably because they’re the ones using and benefitting from the technology. They’re people that couldn’t strong a sentence together before, but are now generating whole articles in minutes. The people who couldn’t tell you the difference between cyan and turquoise are now claiming to be able to generate “amazing” images.
The problem is that while users celebrate these Gen AI advancements, and enjoy benefitting from it, they conveniently forget that artists and authors rely on copyright to protect the value of what they have created. It’s original, it took time, effort, and more importantly skill. It’s their livelihood, and when AI gives everyone the ability to copy what they have created, it dilutes the value of what true artists and authors do.
Even if the AI companies were to pay for the copyright for the millions of books and resources that they’ve used to train their LLMs, it wouldn’t change the ethical debate about copyright. Does payment equal ownership? Even that’s a grey area that goes back centuries.
Throughout the ages, artists and authors were traditionally sponsored by benefactors, so although they may have been the original creators of the works, there was the question of whether the benefactors also had a right to ownership. After all they had paid for the works to be created. This was debated when copyright laws first came into place early in the 18th century.
Judges had to weigh up the economic right of the benefactor, versus the moral right of the artist. Then there was also the industry right to reproduce a work, or the ownership right of the person who bought it. The difference is that copying wasn’t a big deal back then as there were no mass printers and copying something often took as much skill as creating originals.
The same can’t be said of Gen AI in the average user’s hands today. Prompt, click, output produced in seconds. It might impress, maybe even start new trends on social media, but is it really something that could or should be copyrighted?
