Assuming that copyright protection presupposes a creative achievement by a human being, copyright law faces a national and international challenge in view of the use of artificial intelligence to generate images, films and texts. Are such products protected by copyright? And who is the author?
First of all: The law does not adequately answer this question and there is little case law on the subject (in particular no case law from German courts to date).
However, the question has an impact on the activities of artists, designers and authors who use AI as a tool in their work. If the result remains without copyright protection, if it plays no relevant role in the value chain, the author cannot reap the rewards of his work in the same way as is the case with works that were created purely by the author’s brain and manual labour. This is particularly unfortunate when you realise that AI is not only a tool for creators, but also replaces them in some cases, especially in the creation of so-called applied art (e.g. logo development).
In its latest report on copyright protection for AI-generated works, the US Copyright Office addresses the questions of whether a machine can be considered an author and what role humans play in this creative process. The US Copyright Office is the federal agency charged by law with administering US copyright law. In a sense, the US Copyright Office administers the copyright system in the USA. Its core responsibilities include registering copyrights, providing information and advice on copyright, maintaining records of copyrighted works, and assisting Congress and other agencies with copyright issues.
Opinion of the US Copyright Office
The report clarifies that current copyright law only protects works based on human creativity. AI-generated products can only enjoy protection if a human has played a significant creative role in the creation of the product. On the one hand, this means that pure AI products without any human intervention are not eligible for copyright protection. However, if a human being makes creative decisions and edits the AI-generated material or controls its creation, the result can enjoy copyright protection.
The relevant human influence must take place during the use of the AI, i.e. in the creative process of utilisation, for example by editing or arranging the AI results. The development or training of the AI alone does not establish copyright protection for the content subsequently generated by the AI. Human intervention in the creative process of an AI should therefore only be relevant in a significant way at the prompting stage, but not when developing the algorithms and feeding the AI with content.
However, simply entering text commands (prompts) into an AI system should not be sufficient to justify copyright protection. The human contribution must go beyond prompting and manifest itself in a creative selection, arrangement or editing of the AI-generated material. What matters is that the human determines the key creative elements of the work. However, the US Copyright Office emphasises that the human contribution must be assessed on a case-by-case basis. Ultimately, there are no fixed rules as to how detailed or elaborate the prompting must be in order to obtain copyright protection.
To summarise, from the perspective of the US Copyright Office, copyright law only protects AI products with a recognisably human touch. The use of an AI therefore does not exclude copyright protection, but rather establishes copyright if the AI is a tool in the human creative process and human creativity is in the foreground. Purely AI-generated works without human intervention are not protected by copyright.
This has implications for the activities of artists, designers and authors who use AI as a tool. Their work can continue to enjoy copyright protection as long as they have a significant influence on the end product. In practice, this becomes a question of verifiability. Creatives who rely on generative AI must therefore clearly document the human contribution to their work (including the use of AI) and it must be possible to prove that certain human instructions to the AI produced the specific result. This is particularly difficult as long as the AI delivers different results in response to the same prompts. It will therefore probably be necessary to document the entire prompting process, although this is certainly possible with the help of the applications.
Legislative changes not currently necessary
As long as future developments do not give rise to any new problems, the Office does not recommend any changes to the law on the question of whether AI products enjoy copyright protection. The current legal basis (at least in the USA) is flexible enough to take AI-generated content into account.
For us Europeans, the statement from the US Copyright Office has no immediate consequences. It is purely a statement, which is not binding even in the USA, but of course makes the standards transparent by which the Office will assess future applications for registration of copyright-protected works involving AI.
However, it can be assumed that the view from the USA will not be ignored in Europe. Tendencies not to exclude AI products from copyright protection across the board are also recognisable in this country.