While the challenges associated with affording AI-generated works copyright protection is currently a hot topic, the first attempt at a legislative solution was actually made over three decades ago. Noting at the time that it was the first legislature in the world to have considered copyright in the context of AI, and wanting to encourage confidence in AI investment, the UK Parliament created, as part of the Copyright, Designs and Patents Act 1988 (the ‘CDPA’), a category of ‘computer-generated works’, being those generated by a computer “in circumstances such that there is no human author of the work”. Section 9(3) CDPA provides that the author of a computer-generated LDMA (literary, dramatic, musical or artistic) work, “shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken”. This approach of attributing authorship to LDMA works when there is no human author has been followed in a small number of other jurisdictions, but has otherwise remained something of a legislative backwater, receiving only a single judicial mention from the English courts in its 30-year history.
However change is in the wind. Enabled by the convergence of big data, enhanced statistical and pattern recognition techniques, and cheap, readily available processing power, AI has already reached the point where automated content creation is becoming a commercial reality. With this seismic shift in the way potentially protectable works are created, critical examination of section 9(3) is rapidly moving from idle, academic conversation to being a crucial issue for both those in the AI industry and legal practitioners. The issues emerging from this renewed discussion are twofold.
The first relates to the practical application of section 9(3) and the interaction with the traditional concept of authorship. In many situations it will be difficult to draw a clear distinction between works of human authorship (where AI might be a ‘tool’, much like a camera) and computer-generated works where there is no human author. With the boundary between human-authored and computer-generated works is not entirely clear, authorship disputes become increasingly likely, especially as AI-generated works often arise as a result of collaborations between those providing AI software and others providing data to train the AI and setting the AI to a particular task.
The second issue is the requirement that LDMA works must be original in order to obtain copyright protection. The originality requirement is closely linked to the question of authorship, as (aside from computer-generated works) the author is taken to be the person responsible for the protectable elements of the work, ie what makes it original. The relationship between section 9(3) and the originality requirement has not been considered by the English courts and academic commentary varies and broadly splits into three camps. The first focuses on the creative efforts of the persons making the arrangements. The second suggests that there is no originality requirement for computer-generated works, while the third proposes that the originality of computer-generated works should be assessed objectively, ie would the creation of the same work by a human have satisfied the originality requirement?
Lurking beneath the surface of this second debate is perhaps a more fundamental issue: should copyright only reward acts of truly humancognition or does it play a more utilitarian role in society, encouraging the production and distribution of new works irrespective of the manner in which they were created?
The solution may lie in recognizing computer-generated works as deserving of only economic rights akin to those afforded to films, sound recordings, broadcasts and typographical arrangements. These works must not be a copy of a previous work, but need not be ‘original’. The quid pro quo being a narrow protection against only the signal of the work, not its content. However it remains to be seen whether the CDPA does enough to allow this type of solution to be adopted by the courts without the need for further legislation.
Having been drafted in the 1980s, when AI was but a concept, UK copyright law may well need updating to accommodate the realities of AI. For now, however, the debate regarding section 9(3) continues.