Sunday, February 2, 2025

USA-China Tensions Transform Global Market

After the U.S. elections, relations between the...

So-called Artificial Creativity Gains Popularity: Copyright Issues Remain a Challenge

LAWSo-called Artificial Creativity Gains Popularity: Copyright Issues Remain a Challenge

The growing popularity of AI tools has led to the creation of more and more “artworks” produced by algorithms capable of creating music tailored to preferences and moods, painting pictures, and even writing poetry or film scripts. This raises a number of challenges related to copyright issues. There are also doubts concerning the protection of works that AI learns from. Imposing legal bans and restrictions may, however, not be effective.

“Artificial intelligence can replace writers and translators, but it all depends on the results we expect. From a legal perspective, AI is not treated as a creator, therefore, what it generates is not treated as a work. However, the value of the so-called output, or the final effect, depends on the context and that element which—for now and probably always—can only be provided by a human,” says Dr. Iga BaÅ‚os, a lawyer and expert in intellectual property law, in an interview with Newseria. “For me as a literature recipient, it is very important who creates this culture. I like to know the writer’s history, where they come from. We often read works because of our sympathy for the author, because of his style. Therefore, it’s also a question of what our expectations are as a society, what our aesthetic criteria are, and what satisfies us, whether it’s mediocrity or high-level creativity.”

AI-based solutions have already become widely used in so-called artificial creativity, which in turn has raised a number of doubts regarding copyright.

One of the first precedent-setting judgments concerning the legal protection of works created by AI was issued last year by a federal court in Columbia, in a case brought by Stephen Thaler, head of Imagination Engines. Initially, he wanted to register the image “A Recent Entrance to Paradise,” created by artificial intelligence, as a copyrighted work, but the application was rejected by the United States Copyright Office. Thaler appealed the decision in court. However, the court upheld the office’s decision, stating that copyright protection applies only to works created by humans, not by machines that do not have an original concept or legal personality.

Writers, screenwriters, and photographers, in turn, argue that companies developing artificial intelligence illegally train various large language models (LLMs) based on copyrighted content.

“In the United States, there have already been lawsuits filed in courts, but they were dismissed due to a lack of sufficient evidence,” says Dr. Iga BaÅ‚os. “It is difficult because the final product of artificial intelligence usually does not resemble the input works at all. Therefore, it was very difficult for the plaintiffs to demonstrate in court that their works had been used and their copyrights infringed.”

Last year, a lawsuit in such a case was filed against Microsoft and Open AI by the American newspaper “New York Times,” arguing that the algorithms learn from texts published in its columns. Getty Images, on the other hand, filed a lawsuit against the technology company Stability AI, which allegedly used photos without a license to train its Stable Diffusion AI image generation system.

“From our firm’s experience, clients assume that essentially everything they have ever published that has been disseminated on the internet has been used to train AI. The problem here is the verification stage, whether this really took place. Some try, they ask ChatGPT sophisticated, creative questions, indirectly, wanting to check if their work was used for this purpose, if, for example, a fragment that the creator knows best appears in the response. There are still no unequivocal proofs. However, I think the assumption that most publicly available works have already been pulled to train AI is probably correct,” says the lawyer.

According to the expert, imposing legal bans in this area may prove ineffective.

“Any attempts to prohibit the use of works for the purpose of training AI models through legal norms will remain just bans. This does not mean that they will actually be enforced and adhered to by AI model providers. It is much more rational to assume that these contents have been and will be used and to introduce solutions that would compensate for this financial imbalance, or also provide tools to verify which materials were used for training,” suggests Dr. Iga BaÅ‚os.

Copyright issues are also not regulated by the EU AI Act, the Act on Artificial Intelligence, adopted by the European Parliament in 2023. The regulation, which came into effect at the beginning of August last year, is the world’s first attempt at comprehensive legal regulation of AI systems and their applications and limiting the negative effects associated with them. The regulation introduces, among other things, a requirement to inform that a given work, image, or content was generated by AI, which will significantly affect the development of this market in the future.

The confusion around copyrights is not the only problem associated with the rapid development of AI. Another is the threat of disinformation and so-called deep fakes and how users utilize tools based on artificial intelligence.

“Artificial intelligence should not be treated as a threat. It is a tool that—like any other—can be used for very useful purposes. However, it may pose a threat when carrying out tasks for which someone, i.e., an identifiable human, should be held accountable,” adds the intellectual property law expert.

Check out our other content
Related Articles
The Latest Articles