Spurred by the vast potential of artificial intelligence (AI) and machine learning (ML) technologies, many countries are evaluating whether existing legal frameworks for intellectual property (IP) protection are adequate or need to adapt in order to promote key policy objectives concerning AI/ML technologies. Interestingly, not all countries have adopted the same approach. As a comparative example, Table 1 below illustrates how the United States and United Kingdom diverge on copyright protections for works created by AI/ML systems and exemptions to copyright infringement for automated processes using AI/ML systems.
Table 1 – US/UK Copyright Law Comparison Chart
The divergent legal and policy approaches between the United States and the United Kingdom with respect to copyright protection are explored in further detail below, and how those approaches may impact a company’s decision on where to conduct certain business activities and invest in innovative AI/ML technology.
A. Copyright Protection for Computer-Generated Works
The US currently does not recognize copyright protection for computer-generated works without a human author. In fact, the US Copyright Office’s Review Board issued a decision earlier this year addressing whether an artificial intelligence “Creativity Machine” can meet the statutory requirements of an author for copyright purposes. See Decision dated Feb. 14, 2022, available here. The Board held that the Creativity Machine does not meet the statutory requirements of an author, consistent with the Office’s position that an author must be a human being. (Read more about that decision here.)
By contrast, the UK offers copyright protection over computer-generated literary, dramatic, musical, or artistic works without a human author for 50 years. Note that the UK is only one of a handful of countries that gives any copyright protection to creative works solely generated by AI.
B. Copyright Infringement Exemptions for Automated Computational Techniques
The US does not currently offer a blanket copyright “exemption” for entities to use (e.g., make copies) of copyrighted works for automated processes, such as training an AI/ML system. Indeed, making a copy of a copyrighted work for such purposes may constitute prima facie copyright infringement under current US law. The key question, however, is whether such automated processes and training activities (sometimes characterized as “non-expressive” use) may nevertheless be defensible as a fair use under 17 U.S.C. § 107. The fair use defense requires courts to consider four factors:
- the purpose and character of the use, including whether such use is of a commercial nature or is for nonprofit educational purposes;
- the nature of the copyrighted work;
- the amount and substantiality of the portion used in relation to the copyrighted work as a whole; and
- the effect of the use upon the potential market for or value of the copyrighted work.
While certain decisions suggest that copying of works for non-expressive automated processes, such as training an AI system, may constitute a fair use, the issue not fully settled in the US and requires a fact intensive inquiry under the four fair use factors.
By contrast, the UK government recently issued its response to the UK Intellectual Property Office’s Consultation, proposing a new copyright and database right exception that allows text and data mining (TDM), i.e., automated computational techniques used to analyze large amounts of information, for any purpose. Rights holders will no longer be able to charge for UK licenses to TDM, and will not be able to opt-out of the exception. But there will be a requirement for lawful access, and rights holders may accordingly charge for access to a platform where the works are made available. The UK government’s response notes that the proposed exception “make the most of the greater flexibilities following Brexit” and “will help make the UK more competitive as a location for firms doing data mining.”
C. Practical Considerations
There are numerous effects of the divergent copyright laws between the US and UK, including, inter alia:
- Computer-generated literary, dramatic, musical, or artistic works created without a human author may be protectable by copyright in the UK, thereby potentially creating new revenue streams for dissemination (g., licensing) of such works in the UK.
- Computer-generated works created without a human author could fall into the public domain in the US if publicly disclosed, and be freely used by anyone.
- Companies training AI/ML systems in the UK using works protected under UK copyrights can do so without a license so long as the protected works are accessed lawfully.
- Companies training AI/ML systems in the US using works protected under US copyrights (g., images, news articles, songs) may potentially be liable for copyright infringement without an express copyright license and may have to rely on a fair use defense.
Such legal implications may have practical effects as to where an entity decides to conduct certain business activities. For example, companies looking to create new revenue streams from licensing unique computer-generated works may have better success controlling the dissemination and use of those works in the UK. However, the divergent copyright protections over machine-generated works between the UK and US may matter less if an entity instead intends to derive value from its machine-generated works under the trade secrets laws. Moreover, the UK’s proposed express copyright exemption may make the UK a more attractive jurisdiction to engage in data mining and other automated processes, especially when compared to the potential legal uncertainty and threat of litigation in the US.
Stay tuned for the text of the UK’s proposed copyright exception, and most importantly, stay nimble because IP policy and laws are likely to continue to adapt in view of disruptive technologies like AI/ML.