Artificial intelligence (AI), and generative AI in particular, holds great potential as a tool to assist, support and enhance human creativity. How can this positive potential be unlocked and what role does copyright law play in that respect? Last year, we addressed the “input” side of the equation with our Policy paper #15 on using copyrighted works for teaching the machine. In this blog post, we look at the issue of copyright protection of AI-generated outputs.
Copyright protection of outputs was recently the focus of the WIPO Conversation on AI and IP 10th session and featured in the Policy Questionnaire on the Relationship between Generative Artificial Intelligence and Copyright and Related Rights published by the Hungarian Presidency, on the basis of which Communia developed positions on the key questions raised. Here’s a summary of the main points.
The rationale of copyright protection
Copyright exists to incentivize human creativity. Where the requirements of protection — authorship and originality — are not met, protection should not arise. This rationale has been put to the test over the centuries since copyright first became a legal regime. Indeed, with every new technology, copyright’s fundamental principles had to be reaffirmed and applied in new contexts, against new tools, and to support new forms of creativity. The photo camera, mechanical piano, digital technologies, and the Internet have all led to conversations about how copyright can keep up with the evolution of how humans create. Questions at the nexus of generative AI and copyright thus follow a well-trodden path.
Authorship
The notion of human creativity is a bedrock principle of the copyright system. As such, to meet the authorship requirement, AI outputs need to be the creation of a human author. Under EU law, an author “is a natural person, i.e., a human being who, at least to a specific extent, is exercising a subjective judgement in the composition of a work and has control of its execution” (The notion of ‘authorship’ under EU law—who can be an author and what makes one an author? An analysis of the legislative framework and case law).
One thing is clear: direct human, authorial involvement is essential. Purely autonomously generated outputs (without human involvement) should not benefit from copyright protection. Granting exclusive rights for such content would impede copyright’s core purposes and deprive the commons and public domain of new materials, without any recognizable public benefit in the form of bolstered creativity.
Originality
The key issue with originality is whether the outputs originate from the author. Outputs would need to be the “author’s own intellectual creation” and “reflect his personality in that subject matter, as an expression of free and creative choices.” The author must exert sufficient creative control over the expression of ideas in the final output. These conditions have been developed over the years in EU case law, including the following major cases: Brompton, Cofemel v G-Star Raw Infopaq I, Football Dataco v Yahoo! UK Levola, SAS Institute, BSA and Painer.
In their 2020 study for the European Commission and as summarised in a blog post, P. Bernt Hugenholz, João Pedro Quintais, and Daniel Gervais divide the process of using AI to generate outputs into three parts and highlight at each stage whether it is the human or the AI that is “in control:” (1) conception (human); (2) execution (AI); and (3) redaction (human). “Originating” may thus occur, for example, by entering prompts, iterating, refining, editing and finalizing the output.
Originality will need to be evaluated on a case by case basis, by undertaking a holistic analysis of the creative inputs over the course of the entire creative process (prompts together other means of exercising control over the expression of free and creative choices) on the whole process at each of its stages. Non-original elements of the outputs, such as ideas, facts, etc., will need to be excluded from the scope of protection.
Conclusion and next steps
The fundamental purpose of copyright is to incentivize human creativity and promote public access to such creativity. We must thus ensure that it continues to preserve this precious balance in the age of generative AI. We will turn these inchoate thoughts into a policy paper early next year, stay tuned!