What is AI and Copyright Law?
The defines AI as a system capable of carrying out a task 鈥渃onsidered to require human intelligence, with limited or no human intervention.鈥� As Osgoode Hall Law School Professor Carys Craig describes in 鈥�,鈥� the copyright system 鈥渋s the body of law that grants exclusive rights over works of authorship,鈥� which includes the sole right to publish, reproduce, or perform the work. The exercise of attributing authorship 鈥� and subsequently, ownership 鈥� is essential to copyright law. Even if an AI system is the true author of a work, it will not necessarily receive the title of owner, the one who bears the copyright. Indeed, current copyright laws in the United States and Canada prohibit the granting of such a right.
听
The American Context
In February 2022, the of the , tasked with handling copyright applications, refused a claim for a work that was autonomously created by an AI system. In November 2018, Steven Thaler had filed an application to register an artwork, seeking ownership under the as the machine鈥檚 owner. His claim was refused in August 2019 because the work lacked 鈥渢he human authorship necessary to support a copyright claim.鈥� Now, in 2022,upon reconsideration, the Office refused to abandon 鈥渋ts longstanding interpretation of the Copyright Act, Supreme Court, and lower court judicial precedent that a work meets the legal and formal requirements of copyright protection only if it is created by a human author.鈥�
Within section 101 of the , a 鈥渨ork made for hire鈥� is defined as a work produced by an employee within the scope of their employment, or a 鈥渨ork specially ordered or commissioned for use.鈥� However, it requires that the parties expressly agree in writing to such an arrangement. Here, it becomes evident that an AI system is incapable of providing consent or signature. Although the standard to attract copyright protection is not difficult to attain, the barriers that exist for AI systems in U.S. copyright law are significant, and substantially limiting.
听
The Canadian Context
In a 2004 decision, , the Supreme Court of Canada concluded that an original work must 鈥渂e the product of an author鈥檚 exercise of skill and judgment,鈥� and it cannot be characterised as 鈥渁 purely mechanical exercise.鈥� Similarly, the holds that for a design to be considered original, it must 鈥渂e the result of the designer鈥檚 creative endeavor.鈥� Although the Court in CCH Canadian Ltd. did not explicitly refer to AI systems, it is assumed that such systems would not meet the threshold required for originality under Canada鈥檚 .
In 2019, Adam Basanta was subject to a on behalf of artist Amel Chamady. Basanta鈥檚 project used an AI system which was trained to identify patterns that were similar to existing images in a database of contemporary abstract art. However, the parties settled the case, leaving many questions unanswered regarding the place of AI in copyright law. These included, in particular, the following: What is Canada鈥檚 current approach to the granting of copyrights to AI-assisted works? What are the limits in Canadian jurisprudence for the copyrightability of artworks created with the assistance of AI systems?
In July 2021, the Government of Canada launched , whose goal was to gather expert information in anticipation of a review of the Copyright Act. The Consultation stated that, if used ethically, AI has the potential to drive productivity growth in the economy and to encourage investment in transformative technologies.
听
AI鈥檚 Potential Implications for Copyright Law
The Canadian government鈥檚 position 鈥� as demonstrated in their Consultation 鈥� that AI has the capacity to positively affect the economy enables us to question whether the continued denial of copyright protection to works created by AI systems may have the , stifling investment in 鈥� and the creation of 鈥� innovative technologies. However, even if courts and regulatory bodies decide to vest copyright protection in such works, where multiple parties are involved in the creation of the AI system, such as the investor, the creator of the technology, or the one who generated the code.
Although there are varying levels of human intervention and assistance in an AI system鈥檚 work, these nuances have yet to be considered in the Canadian and American contexts. A revision of Canadian and U.S. copyright laws should include the possibility for AI-assisted works to be considered 鈥渙riginal鈥� and attribute authorship to the creator of the system, as difficult as it may be to determine who that legal person is. Because the individual is the one who conceptualizes and directs the development of the work, they should be attributed the right of ownership. This would also increase the potential for investment in AI technologies and promote innovation. For instance, section 9(3) of the UK鈥檚 stipulates that in the case of a work 鈥渨hich is computer-generated, the author shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken.鈥� In this manner, the UK framework may serve .
听
The Case for Copyright Law Reform
An argument can be made that , because systems develop a unique style as they build their databases, thus making it increasingly unlikely that any two machines would create the same work. Nevertheless, any copyright law reform is unlikely to grant the right of ownership to an AI system, precisely because there must always be a legal person to hold the copyright.
Although we may rightly be a and treating them as responsibility-bearing entities, any forthcoming copyright law reform in Canada and the U.S. should allocate the right of ownership to the creator of the AI system. This right will necessarily include a duty to be responsible for the actions of the system. Yet, it appears as if AI systems are 鈥� quite literally 鈥� lawless: they are not subject to constraints regarding reasonable behaviour, nor do they adhere to a code of ethics. As such, even if copyright laws were to be expanded as to grant copyright protection to AI-assisted works, it remains to be determined what level of fault may be required of the creator for the actions of its AI system. Should copyright law apply a regime of negligence, or perhaps one of strict liability?
If copyright law were modified to assign responsibility for AI systems to their creators, this may push the limits of the concepts of foreseeability and fault. We find ourselves in a cyclical dilemma, in that such a rule of strict liability may stifle innovation and investment in AI technologies, which would ultimately have the same effect as a refusal to grant copyright ownership to creators.
Although claims for copyright protection in the works of AI systems have been continuously rejected in favour of allowing copyright laws to function effectively, it is now high time for Canada and the U.S. to substantially reform their copyright laws to adapt to the dilemmas posed in the age of artificial intelligence.
Any copyright law reform should develop out of a recognition that artificial intelligence is highly prevalent in the creation of works in all fields, and that it will undoubtedly continue to be employed for creative purposes. The Canadian Consultation鈥檚 findings, which have yet to be published, will emerge out of the evidence provided by experts in the field of AI, and as such, it would be beneficial for such experts to be involved in the process of copyright law reform. In the Canadian context, the objectives of the government鈥檚 Consultation reveal that a failure to address the place of AI in copyright may be detrimental to innovation and the nation鈥檚 economy.
听