The effect of generative AI on ownership and copyright is the subject of debate and legal challenges due to the uncertainties associated with human involvement and training data.
If you use artificial intelligence (AI) to create something, whose intellectual property is it: you, the AI, or the owner of the content the AI was trained on?
The advent of popular generic AI tools trained on potentially copyrighted material from online sources, usually without the consent of the original creators, has added a new dimension to various forms of art and writing, triggering conversations around this question.
The dangers of AI have been intensely discussed over the past few months and part of the ensuing debate centers on the ownership and authorship of AI-generated content.
Apart from being a new tool in the world of technology, AI has proved to be able to perform tasks that till recently only humans could do. Now that AI is involved in manufacturing, laws that have long protected people and their intellectual property are proving insufficient.
In the legal field, one of the main points of discussion is questions of copyright when it comes to images, texts and other forms of art generated by AI models through human input.
AI-generated art became a relevant controversy after AI-generated art work won Colorado State Fair Art Competition in 2022. The piece was drawn by Midjourney, a generic AI image tool, following prompts from artist Jason Allen.
The victory prompted many angry reactions from artists, who claimed that AI would be the death of creativity and art if AI-generated illustrations were deemed more creative than human work.
While generating the correct signal for the piece demanded hundreds of different signals from Allen – the entire process took more than 80 hours – the AI image was considered by many as not worthy of competing with a human creation.
Apart from the fear of AI taking over the art sector and jeopardizing the jobs of many creative people, the winning of an AI-generated image at an art fair also raised questions about copyright issues. In other words, if the artist only described the art but the AI tool produced it, who owns the rights to the generated image?
Who owns the copyright of AI-generated content?
recently paper Published in the journal Science, the researchers discuss copyright legal challenges in the case of AI-generated works. He pointed out that there are a number of issues that make copyrighting this type of material challenging.
“The way the laws work in America specifically is that we have some laws on the books and then when new technologies come out, it is up to the judges to decide how to apply those laws. Will be done, and that means we need to see that it matters. “It’s such a new technology that a lot of the questions haven’t reached the courts yet,” Robert Mahary, co-author of the study, told EuronewsNext.
It is not the first time that legal systems are faced with questions over how to define ownership of a specific creation, where a new device serves as a medium in its creation.
For example, photography was also an invention that changed our understanding of human creation when it first appeared in the 1800s.
later a court Government Photographers own the rights to the images they create, which has made photography a new form of art in itself.
Perhaps this is the closest the case comes to history that can serve as a guide on what to expect from AI-generated content copyright laws.
EU copyright rules for AI-generated content
Since the emergence of generic AI for public use, the European Union has been playing catch-up mainly in terms of regulations.
While the EU’s AI Act is the closest EU countries have come to regulating the use of AI, specific laws for copyrighting AI-generated content have not yet been established and have so far only been decided on a case-by-case basis. can be settled on that basis.
However, the part of the AI Act that members of the European Parliament recently voted on does address the requirement that companies that deploy AI-generated content exposure any copyrighted material that was used to develop their system.
One of the biggest challenges of copyrighting AI-generated content is the possibility of training AI systems to use copyrighted material, labeling it could be a step in the right direction that could potentially lead to greater awareness of AI. Sophisticated copyright laws will lead to – generated content.
What are the challenges in copyrighting AI-generated content?
The main reasons why copyrighting AI-generated content is challenging are ambiguities about human involvement and intentions, the training data AI tools can use to generate output, and various questions regarding ownership.
“Currently for art, this data is scraped from the web and scraped without asking the artists for their consent, without actually informing the artists that the data is being scraped,” said Mahri.
The data on which generative AI is trained includes many creative works that are protected under copyright laws, and which, in most cases, were included in the training of AI systems without the creators’ knowledge or consent .
“So you use the training data to train this model and then in the last step, you have a trained model and then you can use that to generate new outputs. Now, even the first step, even just taking data and training an AI model can raise copyright issues because now you are turning this art into something new,” he said.
In US copyright law, there exists a notion of ‘fair use’ which basically allows creative work to be based on copyrightable artwork – but it must be so transformative that it differs somewhat from the original. This type of altered work is considered separate from the original work of art and is not subject to copyright infringement.
According to Mahri, there are still many questions about whether an AI tool producing a piece of art that mimics the style of an original work would be considered fair use, and even in which cases it is. , there will still be a strong need to compensate the original artists and protect their work.
Although AI does not always generate responses inspired or based on the work of other artists, training data is hard to detect by the tools used to generate the output.
several lawsuits A case has already been filed against AI image generators whose training data includes copyrighted images.
Beyond the complexities of determining what copyrightable work went into producing a specific piece of AI content, another struggle will be determining how much human involvement is actually sufficient to warrant ownership of the work.
Since the main purpose of copyright laws is to encourage art and its creation by protecting artists and their unique creative ideas, lawmakers may need to consider joint ownership.
“We can expand the copyright law and look at things like joint ownership,” Mahri said.
He continued, “Even though copyright law doesn’t currently protect styles, if you create artwork that is similar to someone else’s, they may have joint ownership.”
Considering that, to date, there is no copyright law that specifically addresses AI-generated content, and the situation is still resolved on a case-by-case basis in court, people should consider What they may need to do is how they’re using generic AI in the process. His creative work to avoid copyright issues later.