A brief guide to AI and copyright

Many works are currently generated through use of artificial intelligence, which raises two important questions: Is it possible to protect these works through copyright and do these works expose the user to claims of copyright infringement? The answer to both these questions is “it depends,” and this is not an evasion, but rather analysis of the legal reality. Here are a number of tips that will assist you in protecting works and avoiding legal claims. 

Development of applications based on AI is currently the hottest field in the technology industry. Hot? It’s on fire. Today practically every company is integrating artificial intelligence into its applications in order to improve service, management, purchasing and product development. 

Alongside its potential, artificial intelligence threatens to eliminate many jobs, and it heralds a huge upheaval in many industries. In the meantime, amazing AI applications are tempting many to use them. Students write their papers using ChatCPT, photographers use AI to edit and shorten image processing, and artists use it to design products and to create artistic works. 

However, using AI applications also involves many challenges. One is related to the question of copyright of the AI products, that is, can these products – such as texts, photos, illustrations, musical works etc. – be considered works that are protected by copyright? 

Before we try to answer this important question, let us remember what we mean when we say “copyright.” The Israeli copyright law defines copyright as an exclusive right, limited in time, to use a work (or a significant part of it) in various manners such as copying, publication, public performance, broadcast, making it available to the public and more. This exclusive right exists for specific kinds of works, and they are literary, artistic, dramatic and musical works and also records (which may be interpreted also as recordings), for which it is required that the work exists in some fixed form (that is, not as an ethereal concept that is preserved only in the mind of the creator). 

It is important to remember that copyright is mainly a transferable asset and the copyright owners can transfer their rights in the work – for free or for payment, in its entirety or in part – in a written agreement. Alternatively, they can grant a license to use the work –  exclusively or non-exclusively, in a limited or unlimited manner. 

Involvement of a human creator – essential to the legal outcome

Let’s return to the key question of this article – can AI products be considered copyright- protected works? The Israeli copyright law determines the kinds of works eligible for copyright protection, and as stated, they are literary, artistic, dramatic, musical and records under one of the following conditions: the work was first published in Israel; at the time of the work’s creation, the creator was an Israeli citizen; or the creator’s regular residence is in Israel. 

The term “citizen” clearly implies that the creator cannot be a machine but must be a human being. But what happens in a situation when a person creates using artificial intelligence? Does it still meet the requirements of the law? 

According to the law, in most cases, the first owner of the copyrighted work is the creator of the work (an exception to this, for example, can be seen in a situation where the work was created by an employee, and then the employer is the first owner). Therefore, with works that have been created with direct guidance or assistance from humans, the ownership depends on the degree of contribution and creativity of the human, the terms of use of the platform, and the factual and contractual framework in which it was created (for example, in the case of a commissioned work). 

In the case of works created by autonomous artificial intelligence, here too, as of today, in most countries these works remain “orphaned”, that is, ownerless, and are not subject to copyright protection, that is, they are free for use (unless noted otherwise in the terms of use of the platform through which the works were created). 

In other words, in answering the question of whether AI products can be considered as copyrighted works, we must distinguish between two situations:

The first is a situation in which the work was created by autonomous artificial intelligence – in such a case, the only human inputs are the initial algorithms and system “rules” that were created and determined from the outset by humans. Beyond this, in this case there is no direct influence of the human creator on the final product created by the artificial intelligence, and as of today, in most countries there is a requirement of human involvement to make the work entitled to copyright protection. 

For example, in the United States, no protection is provided for works that were created automatically and exclusively by artificial intelligence. Moreover, the US Copyright Office recently issued a new requirement according to which there is an obligation to declare the involvement of artificial intelligence in the creation process. In comparison to another field of intellectual property, the field of patents, the Israeli Patent Office has already determined that the term “inventor” is limited to humans only. From all this, it is understood that works that have been created by autonomous artificial intelligence will not be recognized as protected in most countries around the world. 

Currently, there are some proposals to grant different or lesser protection to works of this type, but a court ruling from the District of Columbia, USA, issued a few months ago, explicitly determined that “a human creation is a fundamental requirement for copyright” and that “copyright is granted exclusively to humans.” In doing so, the court rejected the appeal by Dr. Stephen Thaler, owner of an autonomous artificial intelligence system and a serial litigant, who sought recognition of artificial intelligence as a creator. 

The second situation deals with a case in which the work was created through direct guidance or assistance of human beings – in this case, artificial intelligence serves as a tool to achieve a defined objective or result, while the human being contributes significantly to the final work. It can be claimed, for example, that an artificial intelligence system, like a camera, is merely a tool used by the creator to express an original idea in a tangible way. In this case, it is quite possible that these works will be recognized as eligible for copyright protection, depending on the human involvement and the ability to prove it.

This conclusion needs to be taken with a grain of salt, since the analysis brought here is correct to date. We are all aware of the fact that the legal system lags behind the dynamic technological reality and it may well be that the rules will change in the future. 

A creative solution is required for the issue of legal exposure

One of the first important questions arising with the introduction of artificial intelligence into the world of individual users is whether training artificial intelligence using protected works may expose users to lawsuits. Training artificial intelligence, mainly through machine learning (ML), is currently an integral part of development in artificial intelligence ventures.

The Israeli Ministry of Justice determined in its opinion that it is permissible to train artificial intelligence with protected works and that excluding exceptional instances, this does not constitute copyright infringement and is considered “fair use,” but – and this is a big but – this does not affect the legality of the products. In the United States, for example, many claims have been filed by artists for illegal use of their works as part of training artificial intelligence models, and we must continue to follow this in order to see how the U.S. justice system, from which Israeli justice system draws quite a bit, chooses to deal with this issue. 

In most cases, artificial intelligence platforms’ terms of use include conditions according to which they do not bear responsibility related to the products. This means increased legal exposure for users. Here enters Microsoft. Microsoft has invested a lot in artificial intelligence through products such as Copilot, GitHub, and Bing Chat, which can create unique code, text, and pictures on demand. The software giant announced that it will provide legal defense for clients who are sued for copyright infringement for content created using its artificial intelligence. This is undoubtedly a creative solution designed to reduce users’ concerns and encourage the use of Microsoft’s artificial intelligence engines. However, as in any other legal situation, it is very worthwhile to read the fine print and understand the scope and limitations of the defense that Microsoft offers. Will other platforms fall in line with Microsoft’s approach? Time will tell. 

The bottom line: What uses of AI works are permitted? 

So, what uses are permitted for works created by artificial intelligence? Well, it depends on quite a few factors, such as the type of work, its legal status (protected/free for use), and its content. For example, the law regarding a protected work is different from that of a work that is already in the public domain, certainly for the purpose of training artificial intelligence. In parallel, you need to check the platform’s terms of use – does this platform prohibit commercial use? The content of the work is also meaningful. Thus, for example, a photo of a canine astronaut wearing Adidas shoes may be very creative and original, but in the same breath may violate the Adidas trademark and expose you legally. Alternatively, a wonderful painting of a famous person may violate that famous person’s right to publicity. 

Another question that must be asked is whether the work is derived from other protected works. That is, does the wonderful text I created through ChatGPT suffer from over-similitude to other texts that I used as references? All these answers will have an impact on the final conclusion. 

A number of useful tips

In order for you to successfully navigate the AI era, here are a number of useful tips to help you reduce legal exposure and maximize the ability to protect your works:

1. Carefully read the terms of use of AI generator platforms – These are your “rules of the game”. Play according to the rules, and you will be able to significantly reduce legal exposure. Pay particular attention to the question of whether the platform disclaims all responsibility or whether it, like Microsoft, assumes all responsibility, and whether there are limitations on the way that created works can be used. 

2. Take an active part in the creative process – Be as specific and detailed as possible in the guidance given to the AI. It is worthwhile to take an active part in creating by doing your own editing of the initial product, by editing the language and the graphics, editing different elements, integrating other works (of course it is preferable that these be original), etc. 

3. Anchor ownership of products in an agreement—If you have collaborated with others in the work (employees, freelancers, friends, partners), it is highly recommended that you anchor ownership of products in a clear written agreement, which determines, among other things, an obligation to create in accordance with the law and not violate the rights of third parties. 

4. Make sure that you have not used protected content – Pay attention to trademarks, known figures, as well as names, illustrations, and photographs of people – whether famous people (using the image of a famous person may violate their right of publicity) or private individuals (using the image of a private individual may violate their right to privacy). 

5. Create clear work procedures and preventive internal policies in organizations —If the use of AI generators is intended for business purposes, you, as business owners, have the responsibility for your employees’ actions, and therefore, it is very important to create clear work procedures that include your oversight of content created and the way it is created. 

In conclusion, as with any revolution, the artificial intelligence revolution, alongside the challenges and natural upheaval it brings, also creates new opportunities. Certain positions will disappear while new positions will be created. In order to take advantage of the enormous potential inherent in artificial intelligence, we must use it with wisdom and care. 

This article dealt with the question of protecting human creations, but perhaps the most important question is the question of protecting humanity. This sounds a bit dramatic, but this question has occupied many writers, creators and scientists in the last century. In this context, it can be said that the guidelines regarding creation using artificial intelligence are also valid for the general way of using it. If we abandon our human intelligence and surrender completely to autonomous artificial intelligence, we will give up our rights and degenerate. We must use artificial intelligence as a tool and as a means to achieve a goal, while using our intelligence and our personal and human abilities. Only in this way will we be able to grow and develop, and some would even say survive, the artificial intelligence revolution.