Copyright protection in machine-generated works is not a new issue for law makers. The traditional concept of human authorship was first challenged with the emergence of photography and this has continued every time a new technology comes about.
In the U.S., the case of Burrow-Giles Lithographic Co. v. Sarony, 111 U.S. 53 (1884) extended copyright protection to photography. The court found that the photographer Napoleon Sarony had copyright in his picture of Oscar Wilde by setting the scene and exercising control over the subject matter of the photograph.
Years later, in 2005, the U.S. court in Mannion v Coors Brewing, 377 F.Supp.2d 444 found originality in the rendition, the timing and the creation of the subject.
Then, in 2018, the degree of a photographer’s control over the subject matter arose again with the so-called “monkey selfie” case (Naruto v. Slater, No. 16-15469 (9th Cir. 2018)), questioning whether a non-human could be the author of a copyrighted work.
Like many other countries, in the U.S. copyright subsists automatically in an original qualifying work. Copyright registration may be obtained to create a public record of the work and to assist the copyright owner in enforcement of their rights. Copyright registration in the U.S. is obtained by application to the U.S. Copyright Office.
Most recently, in Thaler v Perlmutter (D.D.C Aug 18, 2023), the U.S. Copyright Office refused to register a “wholly computer-generated” image (see our post about this case here). Once again, the Office affirmed its position that “authorship” is synonymous with human creation.
There is, therefore, a significant difference between “AI-assisted” and “AI-generated” works. Whilst human-authored works with a limited degree of artificial intelligence (AI) assistance may still be registrable, works wholly generated by AI technology do not pass the test of human authorship. This is because, arguably, the human author has no control over what an AI program generates, even if the human gives it multiple prompts. The result is most often unforeseeable and random.
On 5 February 2020, the U.S. Copyright Office and the World Intellectual Property Organization (WIPO) held a symposium that looked at the relationship between copyright and AI. Then, on 26 October 2021, the U.S. Copyright Office and the U.S. Patent and Trademark Office held a conference on machine learning and copyright law (see “Copyright Law and Machine Learning for AI: Where Are We and Where Are We Going?”). The conference considered “machine learning in practice, how existing copyright laws apply to the training of artificial intelligence, and what the future may hold in this fast-moving policy space”.
Since then, the U.S. Copyright Office has held multiple events designed to look deeper into the crossover between copyright and AI, at the level of human input required for copyright protection; text and data mining; and the future of AI and copyright policy. It appears that, although the U.S. Copyright Office is, in theory, open to registering works of AI authorship, it needs to be convinced that there has been sufficient human input. So far, the few applicants that have attempted to register works of AI authorship at the U.S. Copyright Office have failed to show sufficient human involvement to allow their registration to proceed.
One case to watch is a second attempt by Kristina Kashtanova to register her AI-generated (or AI-assisted?) artwork. Following a failed attempt to register her graphic novel (when the U.S. Copyright Office subsequently reversed its decision to register her copyright), she has since applied to register a new artwork. This time, Ms Kashtanova sketched out her “Rose Enigma” before feeding that sketch to the AI program Stable Diffusion (see further discussion here). The Office’s decision is currently pending but it will no doubt be interesting to get some clarity over registration of sketches modified by AI.
Against this background, in early 2023, the U.S. Copyright Office launched a study to examine the copyright law and policy issues raised by generative AI. According to information on its website, the study “will collect factual information and policy views relevant to copyright law and policy. The Office will use this information to analyze the current state of the law, identify unresolved issues, and evaluate potential areas for congressional action”.
The need for this initiative came from the Office’s actual experience of examining applications for registration that claim copyright in AI-generated material. The Office has received applications that have either named AI technology as the author or co-author of the work or have included statements indicating that the work was produced by or with the assistance of AI. Other applicants have not disclosed involvement of AI in the creation of the original work and, once the truth came out, had their registration revoked (such as the first registration attempt of Kristina Kashtanova).
The U.S. Copyright Office subsequently published a notice of inquiry in the Federal Register on 20 August 2023. Initial written comments were due on 30 October 2023, but the deadline was then extended until 6 December 2023. For further information see the webpage for the Copyright Office’s AI Initiative.
The U.S. is not the only country to recognise the need for policy guidance on copyright protection of AI-generated works. By contrast, in the UK, AI-generated works can be protected by copyright as “computer-generated” works under section 178 Copyright Designs and Patents Act 1988 (albeit there is no copyright register in the UK). However, there is a concern that the concept of “copyright-generated” works was created before the era of generative AI, and, therefore, requires revision. The UKIPO continues its consultations to keep the current law under review, to maintain the UK’s reputation as the world leader in AI. As such, the UK government has previously announced its plans to introduce a code of practice on copyright and AI (we have previously reported about it here).
In the EU, the use of artificial intelligence will soon be regulated by the brand new AI Act – “the world’s first comprehensive AI law” (see further information here). On 14 June 2023, MEPs adopted Parliament’s negotiating positionon the AI Act. By the end of this year, we may already see the final form of the Act.
To conclude, AI is no longer a mere buzzword. AI-assisted and AI-generated works are becoming ever more prevalent. Although fears of replacement of human creativity with machine algorithms may be premature, it is as important as ever to keep AI regulation under constant review and ensure that law, whilst protecting human authorship, does not stifle innovation and allows free development of new technology.