Copyright and artificial intelligence

Just as humanity went through historical periods in which its intellectual property rights were not protected, today it again faces the risk of having its rights threatened, not by a human being, but by a machine. Artificial Intelligence technologies have reached a level where their use allows the development of works subject to copyright.

With the accelerated advances in artificial intelligence (AI) technology in recent years, the creation of machine-generated content has become a reality. AI is now capable of creating original works, such as text, music, images, and even visual art, which raises a number of legal issues, including the copyright of these works. Who owns the copyright when a work is created by an AI? How does copyright law apply to machine-generated works? What are the challenges and ethical issues associated with copyright protection in AI-authored works?

Artificial intelligence refers to the ability of a computer system to learn and execute tasks that normally require human intelligence and can be divided into two main categories: weak AI and strong AI. Weak AI, also known as narrow AI, is designed to perform specific tasks based on data and programmed algorithms, such as speech recognition, product recommendation, or even self-driving vehicles. On the other hand, strong AI is an AI that has human-equivalent cognitive ability, capable of performing tasks that a human can perform.

There has been a lot of talk lately about Generative Artificial Intelligences. These are the AIs capable of creating new content from a pre-existing set of information. Over the years, the A.I. has been trained with examples of questions and correct and incorrect answers, which means that when it receives a question, it returns the most probable answer, even if it is a simple operation, such as 2 + 2.

With the increasing ability of AIs to learn and mimic the human creative process, the creation of machine-generated content has become a reality. For example, AI algorithms can create texts, poetry, movie scripts, and music, using vast amounts of data to learn the structure, style, and characteristics of a given type of work. In addition, AIs can also generate visual art, such as paintings and sculptures, using machine learning techniques and neural networks.

Copyright is a branch of law that protects creative works of human authorship. Copyright law varies from country to country, but generally gives the author the exclusive right to control the reproduction, distribution, exhibition, public performance, and creation of derivative works of his or her original work. These copyrights are automatic and protect works as soon as they are created in a tangible form, such as when a text is written or a song is recorded. However, the lack of clarity in copyright law on AI-generated works raises a number of ethical and legal questions.

For example, AIs can create works that perfectly mimic an existing style or genre of work, raising the question of whether this constitutes a copyright infringement of the original work. Furthermore, the fact that AIs are trained on data collected from other existing works raises questions about the originality and plagiarism of AI-generated works.

Another major ethical issue is the lack of proper attribution to the true authors of the AI-generated work. This can lead to a lack of real recognition and proper financial reward for those who contributed to the creation of AI and the algorithms that power it.

In addition, the protection of copyright in works generated by AI has its relevance in the protection of the concentrated domain of these rights when it comes to large technology companies. Because tech companies often have access to the data needed to train AIs, they may also have greater control over the creation and distribution of AI-generated content, which can undermine diversity and plurality of voices in the creative industry.

In summary, the copyright of AI-generated works remains a developing field that requires careful attention. Although the legislation around this issue can be complex and vary around the world, it is important that all stakeholders work together to ensure that the creation and distribution of AI-generated content is fair and equitable to all parties involved. The future of the creative industry may depend on it.

In response to these challenges, some jurisdictions have begun to address the issue of authorship of AI-generated works. In the United States, the authorship of works created by artificial intelligence continues to be an evolving issue and there is no specific legislation that addresses the issue. The decision of the Supreme Court of the United States in the case of the monkey selfie, where the authorship of the work was denied on the grounds that it was created by an animal and not by a human being, can be seen as a relevant reference in this context. This decision raises questions about the attribution of authorship to works created by AI, since AI is a non-human entity, with no capacity for intention or consciousness. Recently, the United States Copyright Office (USCO) issued guidance about the content produced by AI that can be protected by Copyright. In the case of images generated 100% by AI, they will not be protected by this right.

Despite the understanding that AI-created works are the result of a human request to a machine, the production of the work itself is carried out by technology. Furthermore, according to the USCO, AI technologies do not allow for creative user control. In this regard, according to the country’s legislation and judicial precedents, copyright can only protect content that is the result of human creativity, the term “author” being present in the Constitution exclusively for humans.

Leadership in the development of artificial intelligence is primarily a goal of China, where there have been two major lawsuits on the subject. In the first, known as Feilin vs. Baidu, there is a trend towards the public domain, as originality is seen as insufficient. It is possible to criticize this decision, since in the absence of explicit legal provisions on the protection of works produced by artificial intelligence systems, these may fall into the public domain, which could discourage the development of new artificial intelligence technologies.

In the second case, Shenzhen Tencent v. Yingxun, the Nanshan District Court decided that a financial article produced by an artificial intelligence system should be protected by copyright. The court considered that the form of expression met the requirements of intellectual creation, since there was human intervention in the selection, analysis and judgment of the information and data relevant to the matter, condemning the defendant Yinxun for violation of copyright.

In Brazil, copyright legislation does not specifically mention works created by artificial intelligence. A bill is being processed in Congress to create rules on the use of artificial intelligence in the country. Approved by the deputies and awaiting a vote in the Senate, Bill 21/2020 creates a legal framework for the use of this type of technology by the public power, companies, entities and individuals. Currently, the Copyright Law (Law No. 9,610/98) establishes that the author is the natural person who created the intellectual work, being the owner of the copyright on it. However, the issue of AI-generated works remains an evolving topic and can involve complex legal and regulatory interpretations.

Some scholars argue that, in these cases, copyright could be attributed to the developer or owner of the AI, who would be considered the author of the work. Others understand that the works created by AI should be considered collective works, since they involve the contribution of various authors, including the programmers, designers and engineers responsible for the creation of the technology.

The great discussion around this hypothesis revolves around the degree of control and predictability that a programmer can have over the result of a developed work, since the greater the autonomy of the AI, the less its interference will be.

Although he is responsible for the AI technology itself, it is not his personal choices that determine the outcome of the work. It is like when a photographer prepares his equipment to photograph a sunset, but the weather is cloudy and the imagined result is not achieved, since the weather cannot be interfered with. The same goes for AI. The programmer writes the algorithm, but the result achieved will reflect something close to what was imagined, since it does not control all the elements that make up the work.

This does not mean that the programmer will not receive compensation for the developed system, in this case the AI, since not only will he receive it, but he will also be able to restrict its use and creation without his authorization.

In the United Kingdom, the first country to grant explicit copyright protection to AI or “computer-generated” works, a law has been in force since 1988 that attributes authorship of works generated by artificial intelligence systems to whoever created these “systems”, which can lead to attribution of authorship to the programmer of the program, even if human creative intervention is minimal. Adopting this strategy will ensure that companies maintain their investments in technology, with the certainty that they will obtain a return on their investment.

In order to protect the intellectual property rights of visual artists in the UK, DACS, an association that represents these artists in the country, is asking the government for measures to ensure that these artists have the right to decide whether they want or not that their works are part of the databases of these applications. According to the organization, authors who choose to make their works available must receive payment when their creations are used, in some way, as part of a new image.

Another important challenge in relation to the copyright of AI-generated works is fair use. In many cases, it is difficult to determine whether the creation of an AI-generated work qualifies as a fair use of a pre-existing work. For example, in several countries, if an AI-generated work is based on an image dataset that is protected by copyright, the creation of the new work will not be seen as a copyright infringement if it is intended to educational use, for research or even for the dissemination of news.

The use of works with copyright for AI training generates economic benefits, harming in a certain way the consumption of the original works, the use of works for AI training not being a possibility of fair use in accordance with the legal system Brazilian. In this case, the use of works for this purpose must be previously authorized by their owners.

Another issue related to the copyright of AI-generated works is the protection of personal data. As AI-generated works are often created from data sets containing personal information, there may be privacy concerns for the individuals involved in these data sets. This can be particularly problematic in areas like healthcare, where data privacy is essential.

It is important to ensure that the algorithms used to create AI-generated works are designed in an ethical and transparent manner. This can help ensure that AI-generated works are fair and unbiased and that they respect the copyrights of other pre-existing works. Since algorithms are created by humans, it is important to ensure that these humans carefully consider the ethical and legal implications of their creations.

Another important aspect to consider is the need to update existing copyright laws to cover AI-generated works. Most existing copyright laws were created at a time when works were created exclusively by human beings. Therefore, these laws may not be adequate to address the specific issues related to AI-generated works. It is important that copyright laws be reviewed and updated to reflect changes in technology and new ways of creating works.

In the European Union, pursuant to an initial agreement, companies implementing generative AI tools will have to disclose any copyrighted material used to develop their systems, which may pave the way for comprehensive laws regulating the technology.

Recently, their parliamentarians met to ask world leaders for an urgent meeting to address the regulation of AI systems, given the speed with which the issue has been developing.

One of the warnings issued concerns the potential for unprecedented misinformation that unregulated AI development can generate, to the point of even replacing humans if left unchecked.

However, by implementing specific copyright registration systems for AI-generated works and adopting ethical and transparent approaches to algorithm design, it is possible to overcome these challenges and ensure that AI-generated works are respectful of copyright and fair to all parties involved.

Among the countries and their respective legislations, what has been observed is that a large part of them determines that the work created by AI must be in the public domain, and cannot be exploited exclusively, nor can its authorship even be claimed. sell these works would not benefit from a scenario like this.

Additionally, the prevailing view is that AI itself cannot be considered an author, as it lacks the creative ability and legal personality to be a copyright holder.

Head of H&A Brazil. Lawyer.